May 17, 2006
JS-4266

Reforms to U.S. Pension Funding and Accounting Rules:
Their Potential Effect on Equity Values and Interest Rates
European Institute’s Sovereign Funds Roundtable
Assistant Secretary Mark J. Warshawsky
London, England

Thank you very much for that kind introduction.  It is certainly a pleasure to be here.

My talk this morning will concern the potential effects on equity values and interest rates of U.S. reforms to funding rules and accounting rules for private defined-benefit (DB) pensions.  As some of you may be aware, each House of the U.S. Congress has recently passed its respective version of a pension reform bill, and it is expected that the bills will be reconciled soon.  It is difficult to say precisely what form the final bill will take, but, relative to current pension funding rules, I expect that reported pension funding shortfalls will more closely track mark-to-market measures and minimum pension funding will make up reported funding shortfalls more quickly.  If this indeed occurs, year to year volatility in equity values and interest rates will flow more directly to pension asset and liability valuation, and therefore to minimum pension funding amounts, than is true under today's funding rules.  And if, as many expect, the Financial Accounting Standards Board's (FASB's) ongoing project to re-evaluate the financial accounting treatment of post-retirement benefits ultimately results in balance sheet measures of pension assets and liabilities that more closely approximate mark-to-market valuations, the flow-through of pensions to corporate earnings will also be more volatile.

Some analysts have concluded that a stronger contemporaneous correlation between pension fund equity values and minimum pension funding and corporate earnings would cause pension fund managers to invest pension funds less heavily in equities and more heavily in fixed income securities that are similar in duration to pension liabilities, thereby sacrificing some expected return for less volatility.  While this conclusion is probably correct in concept and direction, an open question that I think would interest this group is how large the effect would be and whether it would have a significant effect on equity values and interest rates.  That is what I want to discuss today.

But first let me give you a little background on the U.S. private DB Pension system and the reforms being considered.

U.S. Reforms to Pension Funding and Accounting Rules

Funding rules for private DB pensions in the U.S. are seriously inadequate.  While policymakers found it easy to ignore this fact while the stock market was booming in the 1990s, the sudden decline in stock market values in 2000 and the subsequent declines in long-term interest rates used to discount pension fund liabilities made the DB pension system's fragility apparent.  In 2000, underfunded DB pension plans reported total underfunding of $7 billion.  As of 2005, that figure had risen to $450 billion, a more than 60-fold increase.  And there have been record claims on the publicly-administered pension insurance fund.  That insurance fund's actuarial balance declined $33 billion between 2000 and 2005, from a $10 billion surplus to a $23 billion deficit.

Serious pension underfunding is in large part due to funding targets are not based on mark-to-market measures of pension assets and liabilities.   Instead, reported asset values are typically an average of past mark-to-market measures, and reported liabilities are computed using an average of the long-term interest rate over the previous four years.  When long-term interest rates are declining, as they have been since 2000 until recently, reported pension liabilities understate true liabilities.  In addition, plan liabilities are understated because they don't properly account for early retirement and lump-sum payout options.

Not only do current funding rules allow funding targets to move slowly toward their correct values, they also allow measured funding shortfalls to be closed too slowly.  This is especially true for funding shortfalls arising from plan amendments that raise pension benefit levels.  Under current law, these shortfalls are amortized over periods as long as 30 years.  Not surprisingly, many plans have taken advantage of this rule by specifying dollar pension benefit levels in the pension agreement and making frequent amendments raising benefit levels.  A large share of current underfunding is attributable to such plans.

Being a pension specialist, I was eager to help address these problems when I came to the U.S. Treasury in early 2002.  After a great deal of hard work, the Administration put forward a pension reform proposal in early 2005.  I'm proud of that proposal and the key role my office played in shaping it.  The proposal has many important provisions, but I will only mention the three that are directly pertinent to this talk:

1. First, pension funding targets would be based on near mark-to-market measures of pension assets and liabilities.  For purposes of calculating liabilities, discount rates would be gauged to duration in accordance with a high grade corporate bond yield curve computed and published by the Treasury Department.
2. Second, all sources of underfunding would be amortized over seven years.
3. And lastly, for purposes of determining maximum tax-deductible pension contributions, a cushion level of over-funding would be allowed.

The Administration's pension reform proposal served as a starting point for bills passed by both Houses of Congress late last year and that are currently being reconciled.  Unfortunately, both bills are weaker than the Administration proposal, but we in the Administration are now working hard to influence the Congressional deliberations so as to get a stronger bill.  As I said in my introductory remarks, it is hard to predict the precise outcome of those deliberations, but I think we will get a bill that moves us closer to mark-to-market measures of pension underfunding, as well as shorter amortization periods for making up pension funding shortfalls. 

I should back up a moment and mention that everything I've said so far primarily concerns private DB pension plans sponsored by a single employer.  At this time, fundamental market based reform is not being considered for private plans jointly sponsored by many employers in association with a trade union--so-called multiemployer plans.  Fundamental pension funding reform for those plans introduces a host of other issues that must await another day.  In addition, the bills in Congress would not affect employee retirement plans sponsored by state and local governments.  Thus, any direct effect reform would have on the volatility of pension funding would be for private single employer plans only, plans that at the end of 2005 accounted for 35 percent of total DB pension fund assets in the United States.

In addition to stricter pension funding rules for single-employer DB pensions, sponsors of both single and multiemployer plans face the prospect of parallel changes in DB pension accounting rules.  In the U.S., the Securities and Exchange Commission (SEC) has broad powers to prescribe the accounting practices and standards used by companies listed on U.S. stock exchanges, and for the most part it has delegated those powers to the Financial Accounting Standards Board (FASB), a non-government entity.  FASB identifies problem areas of accounting practice, solicits public comment, and issues "statements of financial accounting standards" that are the basis for "Generally Accepted Accounting Principles."

Recently FASB undertook a two-phase project reviewing its accounting guidance for pensions and post-retirement benefits.  Phase 1 has resulted in a draft rule that is now open for comment.  That rule dictates that reported pension liabilities take account of projected benefits as opposed to already-accrued benefits, and that pension assets and liabilities be reported on company balance sheets rather than just in footnotes as is currently the case.  A final rule is expected by year-end.  Phase two of the project, a joint initiative with the International Accounting Standards Board, is more ambitious and is expected to last three years or longer.  That phase will consider advising that companies value pension assets and liabilities at fair market value rather than allowing them smoothing mechanisms, and that any changes in the fair market value of net pension obligations flow more directly to earnings.

From a public policy point of view, the case for making these accounting rules changes is unassailable.  Accounting is valuable only if the numbers have meaning, and the current pension funding numbers have little meaning.  So, while it will likely take a few years, I think FASB will ultimately make these rule changes.

With that background, let me return to my original question:  What effect would these changes in the pension funding and accounting rules have on:  (1) DB pension fund asset composition? and (2) equity values and interest rates?

Effects Pension Reform Might Have on Asset Values and Interest Rates

The changes in pension funding and accounting rules that I have discussed could have effects on financial markets because the new rules could induce fund managers to move pension fund investments away from equities and into fixed income securities, especially bonds with long maturities and durations.  The purpose of such a redirection of assets toward higher duration would be to make fund returns better match the stream of future pension benefits and reduce volatility.

Needless to say, it is unlikely that all or even most of fund investments in equities would be converted to fixed income.  This is because equities earn higher returns than fixed income securities, and over long enough time spans, the risks associated with equity returns are generally reduced.  Plan sponsors would be willing to accept some additional volatility in asset and liability values in the short run to get greater equity returns in the long run.  Therefore, very long-lived liabilities out several decades or more would probably continue to be funded with a substantial proportion of equities.

Moreover, projected future pension benefits arising from future wage levels, mortality trends and so on are uncertain, and so they cannot be completely matched by a selection of fixed income securities even if a very extensive fixed income market is available.  Some pension managers feel that the higher returns from equities are one way of partially making up for this uncertainty.

Nevertheless, pension reform would probably induce some reallocation of pension fund assets from equities to fixed income securities.  An idea of the magnitude of the reallocation can be gotten from survey data provided by the Committee on Investment of Employee Benefit Assets (CIEBA).  CIEBA surveys in 2003 and 2005 suggest that the upper limit to the proportion of DB pension assets that would be redirected from equities to fixed income securities if there were pension funding and accounting reform would be about 20 percent.  At present private DB assets are a bit less than $1.8 trillion, so 20 percent of assets would come to $360 billion.

However, the pension funding and accounting reforms currently under consideration are aimed mainly at single-employer DB private pensions rather than multiemployer.  Single-employer private DB pension assets are around 85 percent of the DB total, which brings the $360 figure down to about $300 billion.

The sum $300 billion is substantial, but in comparison with financial markets in the U.S. overall it is small.  As a fraction of equities outstanding in the U.S., it is about 1.6 percent.  As a fraction of corporate and foreign bonds in the U.S., it is 3.7 percent, but as a fraction of the combined markets of U.S. Treasury securities, securities backed by U.S. agencies and government sponsored enterprises, and corporate and foreign bonds, it is also around 1.6 percent.

And, of course, the rebalancing of pension funds from equities to fixed income securities would not take place all at once.  If it were to occur over, say, two years, the reallocation would amount to a temporary shift of less than 1 percent per year from the equity market to the fixed income market, and after a few years the rebalancing would be finished.

On its face, a 1 percent per year reallocation appears very small.  For example, if the price elasticity for equities is unity, the reallocation would imply an equity price reduction of 1 percent each year, which is certainly an amount that markets should be able to absorb with no lasting effects.  Similarly, the reallocation should not have any sustained effects on the average level of interest rates.

Nevertheless, there might be some short-term adjustment effects in fixed income markets.  This is especially true because the reallocation will probably be directed toward bonds with long maturity and high duration, as pension fund managers extend the duration of assets, and that could cause a flattening of the yield curve at the high end.

One plausible way to get at the magnitude of the yield curve effect is to use analytical work on the relationship of U.S. Federal deficits to Treasury interest rates.  Staff work at the Federal Reserve Board suggests a rule of thumb that a rise of one percentage point in the ratio of the Federal deficit to GDP increases long-term Treasury interest rates by about 25 basis points.^[1]

The $300 billion reallocation from equities to fixed income securities which was calculated above is less than 2‑1/2 percent of GDP.  Distributed over two years, this translates into an increase in fixed income demand of about 1‑1/4 percent of GDP per year, and multiplying by 25 basis points gives a decline in long-term interest rates of around 31 basis points.

However, this is considerably too large because the demand shift toward fixed income will be spread to other bonds besides Treasuries.  Assuming that Treasuries are less than a third of the fixed income market, the resulting temporary effect would be a reduction of about 10 to 15 basis points in fixed income yields at the long end for a few years, after which the effect would be absorbed by markets.

This estimate is very approximate, and another estimate might be gotten from information on Treasury buybacks.  Over a period of about two years from March 2000 through April 2002, the U.S. Treasury conducted a series of buybacks of Treasury securities.  There were 45 buyback operations, and the securities chosen for buyback were in the medium to long maturity range.  The total amount of Treasury securities bought back came to $67‑1/2 billion, which was around 2 percent of the Treasury market at that time.

The buybacks therefore represent an exogenous injection of demand into the long end of the Treasury fixed income market of about 1 percent per year for a couple of years, which is similar to the anticipated move from equities into fixed income arising from pension reform.

Of course, it is difficult to disentangle the buyback effects from the many other effects on Treasury yields over that period, which included a mild recession and substantial monetary easing by the Federal Reserve.  In addition, during that period Treasury announced that the 30-year bond would not be offered anymore, and many market analysts were expecting bigger buybacks than actually were done, so there may have been some exaggeration in the buyback effect initially.

Nevertheless, a rough approximation of the effect of the buybacks can be gotten by looking at the spread of the 30-year Treasury yield to the 10-year Treasury yield.  At least partly in anticipation of the buybacks, this spread fell near the end of 1999 from about 15 basis points to negative levels in 2000, dropping to almost ‑30 basis points around the time of the first buyback.  However, the spread recovered in the next few months, and averaged ‑9 basis points during all of 2000 before moving up to an average of almost 50 basis points in 2001.  These figures are consistent with a temporary yearly effect of about 10 to 15 basis points during the adjustment period, the same as suggested by the deficit analysis, but in the buyback case the effect was bunched in the first year.  Once the buyback operations ceased, there appears to have been no lasting effect on the Treasury market.

These approximate calculations are consistent with the view that any movements in financial markets brought about by pension reform are likely to be temporary, and small enough that they would not be disruptive or destabilizing.  And after a relatively brief adjustment period, the market movements would probably be mostly absorbed with no significant long-term consequences.

It should also be noted that these calculations overstate the expected market effects of pension reform because they ignore the fact that declines in long-term interest rates would encourage an increased supply of long maturity bonds from issuing corporations and other entities.  And in general, if pension reform is done right, it can be expected to increase the productivity and efficiency of allocation of private capital and raise real asset returns and interest rates overall, which would tend to offset the interest rate declines in fixed income markets.

The figures presented so far pertain to private single-employer DB plans, which are the current focus of reform efforts.  However, some observers think that state and local pension funds might also move a share of their assets from equities to fixed income securities because of future funding or accounting requirements.  Assets in state and local pension funds total about $2.7 trillion at present, the bulk of which are for DB plans.

A 20 percent reallocation of state and local pension assets from equities to fixed income would amount to about $540 billion, which is about 3 percent of the equity market and 2.9 percent of the combined fixed income market including Treasuries, securities backed by U.S. agencies and government sponsored enterprises, and corporate and foreign bonds.

At the present time, the possibility of such a market move is highly speculative.  Nonetheless, if it were to occur, it could be expected to take place several years off, probably after the effects of private pension reform have subsided.  And spread over several years, this reallocation would amount to only about a 1 percent move per year for a relatively short time period, similar to the private pension reallocation.  Again, such small disturbances would probably be absorbed by markets. 

I would also note that other changes included in the pension legislation dealing with defined contribution plans, in particular the auto-enrollment and auto-investment provisions, conceivably could have the effect of moving plan participants over time away from fixed-income investments and toward equities.

In contrast to the implications of these estimates of the effects of pension reform on markets, some analysts have compared potential U.S. pension reform to the U.K. pension system, and have suggested that U.S. reform could result in a significantly inverted U.S. Treasury yield curve, similar to the frequently inverted gilt yield curve.

However, this seems very unlikely because there are large differences between proposed U.S. reform and the U.K. system.  The single-standard U.K. pension funding rules found in the Minimum Funding Requirement (MFR) in the Pensions Act 1995 – which was eliminated by the Pensions Act 2004 – and the accounting rules in FRS 17 have strongly encouraged pension funds to invest specifically in gilts.  And the funds chose longer-term gilts so as to lengthen duration in accord with anticipated payments of pension benefits.

As a result of the tendency toward investment in gilts, at present over half of the gilts outstanding are held by U.K. insurance companies and pension funds.  This is very different from the U.S., where such funds hold less than 10 percent of Treasury securities.  Such a big position in gilts by these funds, especially at high durations, may contribute to the inversion frequently seen in the gilt yield curve.

In contrast to the U.K., U.S. reforms stress the use of corporate bond interest rates for discounting future pension benefits, rather than the gilt discount rates emphasized in the U.K. – I have already mentioned the use of a corporate bond yield curve for discounting future benefits.  Consequently, U.S. pension funds desiring to match liabilities with fixed income assets can be expected to include a large measure of corporate bonds rather than Treasuries alone.  Of course, funds will also emphasize corporate bonds simply because they have higher returns than Treasuries.

To summarize these calculations, the U.S. equity and fixed income markets including corporate bonds are very large relative to DB pension funds, and expected movements of pension fund investments from equities to fixed income securities should have minimal effects on U.S. asset values or interest rates.

Are the Implications of Pension Reform for Asset Values and Interest Rates Important For Policy Design?

Now I will put on my policy hat and consider what importance this discussion has for what constitutes good public policy toward pensions.  When we thought through our proposed pension reforms, we understood there could be some implications for asset prices, but our intuition told us they would be relatively small.   

Our motivation for reforming DB pensions is first and foremost a moral one:  Pension sponsors should be made to make good on their pension promises.  I'm proud that the Administrations proposals would make plan defaults and losses to participants less likely and hope that what will ultimately emerge from Congress will move us to this important goal.

Implementing appropriate pension funding policies actually corrects current asset price distortions.  Specifically, the current lax pension funding and accounting rules help pension sponsors shift pension fund investment risk to the publicly-sponsored insurance fund and possibly taxpayers, and amount to a public subsidy for risk-taking.  As with many public subsidies, the result is distorted prices and a misallocation of resources.  By eliminating a price distortion, pension reform enhances the economy's efficiency.

Concluding Comments

Let me sum up by saying that I am cautiously optimistic that Congress will send the President a good pension reform bill for his signature, and that FASB will ultimately recommend economically meaningful pension accounting rules.  If that does indeed occur, we can look forward to steadily improving private DB pension funding, brighter prospects for the publicly-sponsored DB pension insurance fund, and a more secure future for American workers.   

The Administration's pension reform proposals are not rocket science.  They are common sense.  There is only one legitimate measure of pension asset and liability values--what people are willing to pay in the market--and it does not serve the public interest to pretend otherwise.  And if correctly-measured pension underfunding is amortized over a relatively short period of time, underfunding can never get so big that make-up contributions can cause undue financial hardship for employers.  Workers' pensions can be made secure without jeopardizing the viability of businesses. 

I have argued that reforms to DB pension funding and accounting rules would not cause a significant upheaval in financial markets.  The amounts of the shifts in demand across asset classes brought about by the reforms would be very small relative to the overall sizes of equity and fixed income markets, probably less than 1 percent per year for a few years, and would likely be absorbed by markets after some minor temporary effects.

These modest asset value changes are not cause for concern.  In fact, to the extent that reform does cause asset value changes, those changes result from the elimination of a public subsidy for risk-taking.  It can be inferred, therefore, that asset value changes caused by pension reform actually help direct the economy's resources to their highest valued uses.

-30-

See Thomas Laubach, "New Evidence on the Interest Rate Effects of Budget Deficits and Debt," Board of Governors of the Federal Reserve System, May 2003.