The Study in Brief

The problems of employer-sponsored defined-benefit (DB) pension plans in Canada raise two issues: the need for short-run measures to limit the damage; and the need for new pension models to prevent their recurring.

The DB sector's immediate preoccupations are

the result of changes in the economic environment--in particular, a decline in long-term interest rates--that caused their balance sheets to deteriorate, and of changes in accounting standards to more market-based methods that revealed the underfunded state of these plans in stark form.

The immediate policy challenge is to ensure the recovery and/or restructuring of sick plans, and the continued health of sound ones. Extra time and financial scope to work off deficits are good, but current limits on contributions to plans should rise or disappear, while legislation to establish clear title to surpluses for sponsors who must cover deficits is badly needed.

Accounting standards should remain strict, however, to ensure that emerging problems are seen and addressed. It would be a mistake to privilege government-employee plans by relieving them of the same solvency requirements that apply to private-sector plans. Another wrong turn would be resorting to government-sponsored insurance to backstop plans, since this approach creates moral hazards and future liabilities for taxpayers.

In the longer run, policy should sustain and encourage a thriving occupational pension sector that helps individuals save for old age and helps finance the investment that underpins economic growth. But DB plans were in decline long before the recent crisis, and evidence is mounting that the classic single-employer DB plan has fatal agency problems--evident particularly in the tendency for these plans to mismatch assets and liabilities in ways that exposed them to risks far larger than sponsors or participants understood.

Rather than seeking to prop up the classic DB system, then, Canadians would do better to seek alternatives. Existing RRSPs and money-purchase plans tend to impose high decision-making and administrative costs on individuals. A better route would be to promote the development of plans with good features from both models. They could:

* be predominantly money purchase, but with a small and affordable minimum benefit;

* pool investment risk across a large number of individuals at reasonable administrative cost;

* gear contribution rates to a target payout; and

* steer individuals' portfolios toward an asset mix that would insulate them from fluctuating annuity prices as they approach retirement.

Even as Canadians seek to deal with the short-run problems of traditional DB plans, then, they need to develop new models that offer attractive ways to pool resources and save for retirement, while mitigating not only financial risk and longevity risk, but agency risk as well.

In this issue ...

Attempts to shore up the classic single-employer, defined-benefit pension plan are the wrong response to Canada's occupational pension problems. While tax and regulatory changes can help, Canadians need a new approach to retirement income that will pool risks, control costs, and avoid the agency problems that have put many pension promises at risk.

Employer-sponsored pension plans in Canada face an uncertain future. The value of the assets in many defined-benefit (DB) plans falls short of their obligations, and the benefits those plans promise are less secure than they once appeared. Much discussion of this issue takes for granted two policy priorities: repairing and/or restructuring damaged plans; and ensuring that the classic (single-employer) DB model continues as Canada's favoured approach to employer-sponsored pensions.

This Commentary accepts, with some reservations, the first of these propositions, but takes a highly skeptical view of the second. After a brief overview of Canada's pension system, it describes the current difficulties of DB plans and discusses some dos and don'ts for addressing them. It follows other recent observers in urging changes to features of Canada's tax code that impede the accumulation of surpluses above a certain limit, normally 10 percent of liabilities. It also recommends clarifying and strengthening sponsors' property rights in surpluses to remove disincentives to their accumulation--disincentives that not only contributed to current problems but make their recurrence more likely. At the same time, it supports accounting and reporting practices that make fluctuations in pension-plan balance sheets easier for all concerned to track.

This Commentary further suggests that current problems did not result merely from an unhappy series of financial accidents, but are symptoms of defects in the DB model itself. Hence, it concludes that the apparent ongoing decline in DB plans, particularly in the private sector, and the concomitant growth of money-purchase and registered retirement saving schemes, is not the deplorable development many commentators assume. (1) Rather, it a necessary prelude to a more thorough overhaul of employer-sponsored pensions in Canada.

Canada's Pension System and the Problems in its Private Sector

By international and historical standards, the publicly funded pillars of Canada's pension system are in good shape.

The first pillar consists of the Old Age Security and the Guaranteed Income Supplement, which provide basic support for those without significant other income in old age. The federal government runs them on a pay-as-you-go basis. Their benefits--about $13,350 annually for a single person with no other income in early 2007, and $21,650 for a couple--are modest and, being indexed to consumer price inflation rather than wages, should prove manageable for the federal treasury over the long haul (Robson 2006). They are both subject to income-related clawbacks, however: 15 percent for the OAS between incomes of about $61,000 and $100,000, and 50 percent for the GIS for all non-OAS income. Combined with ordinary income taxes and other geared-to-income benefits for seniors, these clawbacks create high effective marginal tax rates that discourage work and saving among low- and middle-income households. (2)

The second pillar is the Canada Pension Plan (CPP) and its Quebec counterpart (the QPP). These compulsory, contributory defined-benefit programs cover essentially all paid and self-employed workers: about 90 percent of the labour force contributes to them in a given year. The CPP and QPP pay maximum annual retirement benefits of about $10,360 a year. A reform package in the late 1990s raised C/QPP contribution rates to 9.9 percent of covered earnings, a level intended to be sustainable for decades. Since they have an accumulated unfunded obligation to older participants, these plans, like social security programs nearly everywhere, require new participants to pay more than an actuarially fair rate for their benefits. However, prudent investment of the funds now accumulating in them will make their long-term viability a function of the performance of the economy as well as of the taxing powers of governments.

Canada's pension system also has a substantial private voluntary/contractual component. The bulk of Canada's labour force participates in employer-sponsored Registered Pension Plans (RPPs) and/or Registered Retirement Savings Plans (RRSPs) on either an individual or group basis. Roughly one-third of all workers belong to RPPs, and slightly more than that have RRSPs (there is some overlap between the two); among tax filers reporting incomes of $60,000, more than 90 percent of employees participate in RPPs and/or RRSPs. (3) The voluntary-contractual component of the system is the focus of much current anxiety. In particular, DB plans--which covered some 95 percent of public-sector and 75 percent of private-sector RPP members as late as 2005 (Horner 2007, Fig. 3.9)--are covering fewer workers. Their sponsors account for a declining share of total employment, and few new plans are starting up.

The Difficulties of Defined-Benefit Pension Plans

The essential feature of DB pension plans is that sponsors promise members a given pension upon retirement--sometimes a specified dollar amount; more usually an amount linked to pre-retirement earnings. In principle, therefore, RPP members can be certain about their incomes in retirement, and are spared the planning forced on people who must provide for themselves. This feature makes many commentators like DB arrangements. They contrast them favourably to the defined-contribution (DC) alternative, under which sponsors make payments on members' behalf into a fund, and members bear the risk that the pension this fund yields will be less (or more) than they expect. And they also prefer them to RRSPs, whether group or individual, whose holders also bear investment risk, and may choose how much, or little, to contribute as well. Notwithstanding these judgments, however, the underfunded state of many DB plans in Canada means that the retirement incomes of their members are less secure than they once seemed.

The obligations of most DB plans, furthermore, are liabilities of their sponsors, many of which are entities with publicly traded shares and debt obligations. To the extent that the pension-plan troubles of particular companies or governments are not fully reflected in share and bond prices, investors may not allocate their saving efficiently. Mispricing of this kind would not simply be an issue for individual entities and investors, moreover. Were problems in one sponsor's pension plan not properly reflected in the price of its securities, plans of other entities holding those securities would appear sounder than they are, and the financial obligations of those entities would tend to be over-valued--a pyramid of mispricing that raises a prospect of systemic risks. (4)

Recent surveys (Armstrong 2006) show that Canada's DB pension plans are in better shape than they were in 2003, largely thanks to the recovery of equity markets. A longer perspective, however, shows that DB plans as an approach to pensions are under pressure. The fraction of workers in any kind of pension plan covered by DB arrangements slid sharply during the 1990s, before the onset of the recent concerns. (5) Money-purchase plans are becoming more important (OSFI 2005, 44-45), not simply because old DB plans are being wound up, but because so few new ones are being created. These longer run trends suggest that Canadians need to ask whether the DB model will figure as prominently in the contractual component of their pension system in the future as in the past, and whether policy should respond either way.

The Immediate Problem: Underfunding

The scope of the sector's most immediate problem, underfunding, is well-documented by regulatory authorities at the provincial and federal levels. The Financial Services Commission of Ontario (FSCO) is the largest single regulator of private pension plans in Canada: about 40 percent of the total. In 2004, DB arrangements covered 90 percent of the 3.3 million members and beneficiaries of the more than 6,000 plans under its jurisdiction. A representative sample of DB plans reporting to it between July 2001 and June 2004 gives a sense of the scale of the underfunding problem at that time. Of 1,718 plans in the sample, covering just under 1.4 million members and beneficiaries, two-thirds (1,167) were underfunded on a solvency basis, more than half (929) were underfunded on a going-concern basis (an issue pursued further in the next section), and nearly 3 in 10 (500) were underfunded by more than 10 percent. (6) At the federal level, data from the Office of the Superintendent of Financial Institutions (OSFI) tell a similar story. At the end of 2004, half the 344 federally regulated private-sector DB plans under OSFI supervision were in deficit, a little more than a quarter by more than 10 percent. Fifty-two of these plans were on a "watch list," indicating some degree of financial difficulty.

More recent information gives mixed signals about the extent to which the situation has improved. By the end of 2005, the proportion of the plans in deficit under OSFI's jurisdiction had risen to 78 percent, and the number on its watch list to 84. (7) A study by Mercer Human Resources Consulting for the Bank of Canada showed a similar trend in a sample of federally and provincially regulated pension plans. The share of plans in deficit rose from 71 to 78 percent between the end of 2003 and mid-2006 (Armstrong 2006, Table 1). There was also a more reassuring finding that suggests that the experience of smaller and larger plans over this period may have differed. The share of total obligations in the sample that was in insolvent plans went from 83 percent at the end of 2003 and 86 percent at the end of 2005 to 49 percent at the end of May 2006. (8) The most recent reading from the representative plans modeled in Watson Wyatt's "Pension Barometer" suggest that a leveling off in the value of liabilities and rising assets will have further improved the position of most plans after mid-2006. (9) From a macro-economic perspective, then, the situation may have improved. From the perspective of individual firms and employees worrying about benefit security, however, the sector is still in delicate condition.

The Immediate Causes of the Problem and Policies to Mitigate It

Until the turn of the millennium, the difficulties with Canada's DB plans were mainly manifested in declining coverage and underfunding problems in industries such as steel or airlines with large pension obligations. Since then, however, four factors combined to produce the solvency problems just discussed and to prompt talk of a generalized "crisis" in the DB sector:

(1) a continuing increase in life expectancy that surprised many;

(2) the end of the long post-1987 equity-market boom in the 2000 crash, from which share prices took a long time to recover;

(3) the onset and persistence of low nominal interest rates, particularly at longer terms; and,

(4) a change in the accounting conventions applied to pension plans" balance sheets involving a shift to "mark-to-market" valuations.

The first of these increased the size of plans' expected pay-outs, the second prompted a reassessment of often extravagant rate-of-return assumptions on a key component of plan assets, the third drove up the present value of plan liabilities by more than it increased the value of their bond holdings, and the fourth brought these effects into sharp focus. Despite the signs of recovery just reviewed, significant risks remain within the system. While a "baseline" projection for 2010 for the sample studied by Mercer has the ratio of assets in deficit plans to total assets in the sample falling to 6 percent, and to zero in the best case considered, a worst-case projection takes this ratio up to no less than 94 percent (Armstrong 2006, Table 4).

Accounting issues

New accounting standards in 1999 that made market-based valuations more important in assessing pension plan's obligations are both a consequence and a cause of recent concerns. Amending them to bury the risks they have exposed, however, would amount to shooting the messenger. Although one might wish, in retrospect, that new rules for addressing deficits had accompanied the rules that made deficits more evident, the changes themselves caused neither the unsatisfactory state of plan funding several years ago nor its subsequent deterioration. The assessment of a pension plan's condition as a going concern, which needs to match its income and its expenditures for decades into the future, requires professional judgement. However, too much discretion in judgements about the financial condition of a pension plan is dangerous, particularly as regards judgements about solvency.

An assessment of a pension plan's solvency answers a specific question: whether, if the plan wound up today and liquidated its assets, it could meet its obligations. Two key principles should apply in answering the question. On the liabilities side, estimates of the present value of future obligations should use a discount rate based on current market data, since such a rate would determine the cash payments due to members were the plan indeed wound up today. (10) Assets, moreover, should be valued at current market prices, which would determine the amount fund managers could raise if they sold the assets today.

Such "marking to market" is not always straightforward. Securities that are widely traded in deep, liquid markets are easier to value than a single equity stake so large that its sale would influence the stock's market price, or a real asset such as a shopping mall or an office building. But the occasional need for judgement in marking to market in the case of such assets does not justify also using, in solvency valuations, smoothed or forecast values of readily observable variables such as interest rates and the prices of widely traded securities. Nor can one justify smoothing and forecasting with reference to the "misleading" volatility in the bottom line that marking to market produces--volatility is a fact of economic life, of which pension-fund managers must be aware, and to which they must respond.

Critics of the marking to market required in solvency valuations are nevertheless right to observe that the typical pension fund is not perpetually on the brink of closing down. Finding a solvency deficit does not mean that a plan must, or ought to, wind up, or that the deficit must immediately be made good. The typical plan is an ongoing operation that must plan over a long time horizon, and an evaluation of plan status from that perspective--the going-concern valuation--attempts to judge whether it is on track to meet its obligations when they are expected to arise. When a solvency valuation reveals trouble, going-concern techniques can help assess its seriousness and design a remedy.

Estimates of the likely evolution of obligations and future investment returns are the essence of going-concern assessments. Some speculative assumptions and projections that would be inappropriate in a solvency valuation are inevitable in that exercise. While the principles underlying going-concern assessments must be precise enough to anchor discussions between sponsors and regulators in the event of disagreement about how far the bounds of reasonableness stretch, they should leave room for the reasonable exercise of expert judgement. Yet for these reasons, going-concern assessments should always be viewed with a little scepticism. The line between reasonable assumptions and wishful thinking is blurry at the best of times, and can shift suddenly. Just such a series of shifts, and a failure of managers and regulators to react to them quickly enough, helped to trigger some of the problems the DB sector of Canada's private pension system still faces.

The Shrinking Equity Premium and Falling Interest Rates

The dominant obligation of pension funds is to provide fixed money-income streams to their members, and the defining characteristic of bonds is that they yield a fixed money-income stream. Although wage growth (which will help determine the value of pensions not yet in pay) is only imperfectly correlated with the returns on bonds, and is indeed correlated to some degree with the return on equities, holding a portfolio that is mainly or even exclusively bonds enables a fund to achieve a reasonable match between its assets and liabilities. A good benchmark for judging the riskiness of a pension plan's balance sheet is, therefore, a portfolio of fixed-income, low default-risk securities that generates cashflows of the right size and timing to meet the plan's obligations. Such a portfolio would match assets to liabilities so that, for example, a shift in interest rates would affect the present value of the obligations and the market value of the assets in the same way.

A common view among managers of DB plans is that matching assets and liabilities in this way is too expensive: the relatively modest returns on high-quality, fixed-income securities would make the contributions needed to finance the plan unattractively high. To make contributions more affordable, the thinking goes, pension plans must take on balance-sheet risk. The most obvious and convenient way of doing so is to invest in company shares. For decades, investment management has held tightly to a view--dubbed by Ambachtsheer (2004, p. 1, fn 1) "an article of faith in the pension investments world of the 1990s"--that a long-term investor in equities who can ride out the short-term fluctuations that make them more risky will reap higher returns. A substantial academic literature lies behind this premise, stemming from a seminal study by Rajnish Mehra and Edward Prescott (1985), which documented an apparently well-established, long-run average difference of just over 6 percent in the rate of return on equities over that on treasury bills in US markets. Further investigations suggested that this excess premium seemed to be a general phenomenon, rather than being US specific, and, crucially for pension fund issues, that it was also relevant to the comparison between equities and bonds. (11)

Whether such an equity premium is available to investors today is a question on which we hesitate to offer an opinion. (12) There are many reasons for doubting that the prospective premium is large--not the least of which is the fact that the widespread belief in an equity premium should mean that share prices already reflect that expectation. What is clear from Figure 1 is that, in Canada, variations in returns between shares and either treasury bills or bonds have been considerable. While total compounded nominal returns in Canadian equity markets were substantially above total compounded returns in bond and bill markets from the 1940s through to the early 1970s, the subsequent picture was much more mixed. Indeed, bonds provided superior returns, on average, in the 1980s and through the mid-1990s, and even bills did better for a short period in the early 1990s.

[FIGURE 1 OMITTED]

Notwithstanding this mixed picture, equity markets moved well ahead of bills again in the late 1990s, and began to outpace bonds again, so the idea that they could be relied on to yield superior returns in the long run retained its influence among portfolio managers. According to Anderson (2005, Table 3, p. 4) 45 percent of the total assets of Canadian pension funds were bonds in 1990, a figure that had fallen to 27 percent by 2000, while equities and other investments moved from 55 to 75 percent over the same period. (13)

The funds' shift out of bonds in the 1990s increased their balance sheet mismatch, and the movements of markets after 1999 revealed the hazards of doing so. While stock market fluctuations garnered most of the headlines, the fall in long-term interest rates was even more important in undermining the balance sheets of Canadian DB pension plans in the early 2000s. The yield on the Government of Canada's benchmark 10-year bond stood at around 6.5 percent in early 1998; after that, its trend was downwards, to 4.6 percent in January 2004, and to around 4.3 percent at the time of writing. Actuaries, accustomed to discounting pension liabilities at 7 percent or more, had to change to a market-related discount rate just as long-term interest rates were hitting their lowest point in more than a generation. The boost this gave to the value of plan obligations was, as a consequence of their asset-liability mismatch, nowhere near the boost it gave to the value of plan assets. (14)

One might argue that the plight of Canada's DB pension plans is, in part, a consequence of the sea change in macroeconomic conditions in the mid-1990s. In Canada, once the 1995 federal budget had brought the country's medium-term fiscal plans into line with a 2 percent inflation target for monetary policy, interest rates began to fall, reducing governments' debt service costs, creating more room for interest rates to fall, and so on. While those specific circumstances do help explain recent developments, they do not tell us why DB plans were so exposed to such an event. An exploration of that topic suggests that the problems of the classic DB model run deeper, and that more than temporary approaches are needed to deal with them.

The Asymmetric Effects of Taxation and the Law

The tendency for pension plans, while in pursuit of extra returns, to invest in assets that are not matched to their liabilities has another consequence. Market fluctuations will occasionally put them in the red. That goes with the territory. Managers who wish their plans to be fully funded (or perhaps in modest surplus) on average should accumulate surpluses in good times and tolerate deficits in bad, to whatever extent seems appropriate to their circumstances. But several tax, regulatory and legal provisions make this approach difficult in practice. It is widely believed that asymmetries in the treatment of surpluses and deficits discouraged managers from accumulating surpluses in the 1990s that might have enabled them to weather the bad years that followed, and that those asymmetries continue to hamper recovery. (15)

First, the federal income-tax code prohibits sponsors of ordinary DB pensions from contributing to their plans when assets are more than 110 percent of plan liabilities. The intent of this rule is to prevent firms sheltering profits from taxation by putting them in their pension plans. We are not convinced that the problem this rule seeks to address is a real one, but if it is, the limit it sets should surely be high enough to allow for the variability that affects pension fund balance sheets over time. A 10 percent limit--a round number that suggests an arbitrary choice--is too small. A degree of asset-liability mismatch that, in other respects, is well within accepted practice will produce fluctuations outside a range of 10 percentage points of balance on a regular basis, so forbidding contributions whenever a plan has a surplus of more than that amount will force plans, on average, to run in deficit.

The legal status of Canadian pension funds also works against their prudent management. Seen from the standpoint of a labour market contract, DB pension rights are deferred compensation the employer must pay when the time comes. While beneficiaries of such a contract have an interest in seeing that obligation secured, their entitlement is to the payments, not the assets that secure them. The language in which many early pension plans were crafted, along with various court rulings, however, have created for them a legal status essentially equivalent to that of a "classic trust" that sponsors must manage for the benefit of members. This means that though a plan's sponsor must meet its pension obligations even if its fund is in deficit--entirely in keeping with treating a DB pension as a form of deferred compensation--plan members also have certain property rights in any surplus the fund may from time to time accumulate. Not surprisingly, sponsors hesitate to accumulate surpluses to which they may lose title, and since the normal course of events will see well-funded plans swing between surpluses and deficits, this hesitation also tends, on average, to produce underfunded plans.

Coping with the Immediate Problem

Certain policy "dos and don'ts" follow immediately from the foregoing discussion.

The first objective of policy must be to help currently healthy plans stay healthy, recovering ones to recover, and badly damaged plans to wind up with as little damage as is practical. Removing the asymmetries in the treatment of deficits and surpluses is vital to this task. Raising or removing the 10 percent upper limit on accumulated surpluses would require a straightforward change to the Income Tax Act. Changing the legal standing of pension funds, however, will be more difficult. Existing case law is what it is, so new legislation needs to establish that pension obligations are what they are in economic fact: namely, deferred compensation agreements, under which pension obligations are contractually binding on employers that enter into them, and under which employers should not face the incentive implicit in the current assignment of property rights to under-provide for them. Action on the part of both federal and provincial governments is needed to remedy this situation, and the sooner it is taken the better. (16) Until it is, pension plans under pressure to fund solvency deficits will reasonably press governments to make the necessary tax and regulatory changes that would let them put their contributions in special trust accounts where they will not be "trapped"--an interim step that has considerable merit.

Along with these two "dos" comes an important "don't." Resisting the current tendency to mark other parts of pension-plan balance sheets to market--let alone tinkering with already established standards to loosen their requirements--makes little sense. Accounting standards exist, among other reasons, to ensure that when problems develop, managers, shareholders and others will see them. A striking feature of the returns on different asset classes shown in Figure 1 is the dispersion of returns around their central tendencies. Figure 2 highlights this dispersion in the specific salient case of the difference between equity and bond returns. This figure shows the remarkable variation in the difference between the compound annual return on equities and the compound annual return on bonds over 10-year periods. It also highlights the difference between the experience before 1970--the period that dominated the early literature on the equity premium--and experience after that date, when the average gap between equity and bond returns has been smaller, and the relative number of instances of a negative gap has been larger. Even historical experience, then, does not unambiguously demonstrate the existence of a reliably large equity premium -certainly not one that can be anticipated, and even booked in advance, in the future. The immediate implication of the undoubted fact that the typical pension plan will not be wound up today is not that accounting rules should permit sponsors to hide problems in the hope they will go away on their own. It is that sponsors should be given sufficient latitude in time and in financial maneuvering to deal with them.

[FIGURE 2 OMITTED]

Seen in this context, the regulations posted by the federal Department of Finance on November 7, 2006, are a welcome and overdue complement to the tightening up of solvency assessment standards that has already occurred. These regulations will let plan sponsors amortize consolidated solvency deficits over five years, extend the period for solvency funding payments to 10 years, provided that no more than one-third of plan members object, and use letters of credit to make up the annual difference between the 5-year and 10-year payments.

This longer amortization period appears reasonable, and is consistent with measures already in place in many provinces. It nevertheless challenges regulators to distinguish situations where sponsors have legitimate workout plans from situations where they are hoping for a miracle. It is certainly possible that new asset categories and hedging techniques might improve the trade-off between long-term returns and risks in pension fund portfolios It is also possible that longterm interest rates will be higher on average over the next decade or so than they are now. But managers who are basing their plans for the future on these assumptions, not to mention their regulators, should keep a careful eye on how the investment strategies that rely on them are working out.

Plans that promise pensions indexed to inflation face a particular and peculiar problem. In Ontario, obligations created by indexing are simply not part of a solvency valuation--a peculiar omission of an element that, where it exists, will typically constitute a large fraction of a plan's liabilities. Where these obligations are part of the valuation, current standards require sponsors to discount their indexed liabilities using the yield on the federal government's real-return bond. This requirement respects the important principle of marking to market, thus ruling out assumptions that might, conveniently, put a plan's obligations in a less difficult light. Yet sponsors obliged to use this rate can legitimately object that it is unreasonable for two related reasons: first, there is so much demand for these bonds relative to their very limited supply that an attempt to offload any significant amount of indexed obligations would move the price; and, second, the yield in this thin market is already peculiarly low. (17)

Since any pension plan with obligations linked to nominal wages has an element of indexation, even if the payments themselves are specified in nominal terms, a key requirement for better asset-liability matching in the long term is a greater supply of assets geared to real income growth and inflation. (18) Meanwhile, for as long as the supply of real-return bonds remains so low relative to demand for them, a more appropriate discount rate for indexed liabilities might be a real interest rate calculated by subtracting the Bank of Canada's 2 percent inflation target from the same nominal-bond rate used in discounting other liabilities. Valuations using that discount rate would need caveats in reports by regulators and in financial statements, since they would widen the gap between the stated price of an obligation and the actual price at which it could be traded in the market. A respectable case can be ruade that the added understanding such caveats would promote among readers of these reports would have important benefits of its own.

Allowing sponsors to use letters of credit to supplement their own payments into plans also brings a useful new element--the financial condition of the sponsor, considered separately from that of the pension plan--into the picture. The creditworthiness of plan sponsors is not, at present, a formal consideration for regulators, and although making it such has compelling logic, it raises some important problems. In particular, public-sector regulators might be inclined to see public-sector plans, which are in theory supported by the general taxing power of the state, as more secure in this respect than private-sector plans. The judgement that government credit is better than private-sector credit is familiar, but is not without counterexamples. It would be unfortunate were it to help create a more permissive regulatory regime for public-sector employers than private-sector ones. The independent judgement of credit quality involved when private lenders provide letters of credit is less problematic from this point of view. Letters of credit do not appear likely to become a predominant method of funding plans. While balance-sheet improvements resulting from this source are not subject to litigation as regular contributions to a trust-like account might be, they trigger no tax deduction as contributions do, and the financial consequences of calling them are so severe that they have a last-resort feel to plan sponsors. Where they are used, however, the independent private-sector assessment of creditworthiness that comes with them provides useful information about the reliability of a given set of pension promises.

Other regulatory changes could make it easier to deal with plans in trouble. At present, federal pension legislation, like that in most provinces, requires that a plan be declared insolvent before its administrators can be replaced. British Columbia and Alberta, by contrast, allow replacement of administrators without such a declaration, providing something short of a "nuclear option" for dealing with a bad situation. This extra degree of flexibility might be useful in other provinces, as might more discretionary power to intervene in the management of funds that regulators deem badly run, or to force sponsors who manage plans directly to cede day-to-day control to a third party.

Occupational Pensions in the Longer Term

Beyond the case-by-case action required by sponsors and regulators to meet the immediate problems facing Canada's private DB pension plans, some larger issues loom. Ultimately, pension funds that cannot now meet their commitments and rail to get lucky enough, fast enough, face unpalatable choices. Higher contribution rates for sponsors and perhaps members, not to mention lower future (or even current) benefits, loom for some. In many cases, hindsight will doubtless show that the parties involved would do better to turn to a suitable financial intermediary, such as an insurer, to transform a problematic DB plan into a set of better secured promises, than to wait until litigation further dissipates the plan's inadequate assets. Our sense is that such options, however unpalatable to many participants and observers, need serious consideration by sponsors, employees and regulators--as do more fundamental reforms.

The fact that problems with DB plans have been so widespread in Canada, and the further fact that other countries--notably the United States and the United Kingdom--have similar difficulties, suggests that classic single-employer DB plans may have fundamental flaws that transcend Canada's particular setup. If this is so, desirable measures such as removing legal and fax obstacles to their efficient management, and deploying regulatory ingenuity to help them with their present troubles, may not be enough. The long-term decline of employer-sponsored DB plans in Canada that began a decade before the onset of their "crisis," and has continued since, might better be welcomed than deplored. It should spur Canadians concerned with the voluntary/contractual part of the country's pension system to look for better alternatives.

Managing and Sharing Risks

Some observers have argued that Canada's employer-sponsored DB pension plans suffer from profound managerial problems. One difficulty is the widespread--though by no means universal--practice of dividing responsibility for their balance sheets. Investment professionals manage their assets. But human resources departments manage the employment contracts that determine their liabilities. So no single person or body deals with the overall asset-liability picture, and in particular, with the mismatches that have rendered so many plans vulnerable to the asset-price and interest-rate movements of recent years, and will continue to expose them to such risks in future. (19) The difficulties of valuing deferred compensation have also given rise to gray areas where the line between professional judgement and wishful thinking is unclear. And the frequency of litigation over pension assets shows how different the understanding of different parties to these agreements has been.

The catalogue of problems is long enough to suggest that the widespread distress in the DB sector reflects more profound agency problems--with representatives of employers and employees making arrangements that suit their short-term purposes but are not in the longer-term best interests of the shareholders and workers on whose behalf they were acting. Indeed, no survey of DB pensions would be complete if it did not ask whether it makes sense for private employers to bear and manage the risks inherent in DB plans in the first place.

The desirability of a voluntary/contractual element in Canada's larger system of retirement income is not, as far as we are concerned, in question. No market economy should rely too heavily on a publicly provided, pay-as-you-go pension scheme to support its aged population. Such arrangements reduce incentives to save, and hence deprive the economy of resources to fund the investments it needs for economic growth. Further, the greater the role of a government-run, partially funded, DB program (on the C/QPP model) in channeling private savings to would-be investors, the likelier is state-sponsored corporatism to intrude into economic life, bringing with it the potential for political stress and economic inefficiency.

The case is strong for a substantial and competitive private-sector presence in the provision of pensions. It does not, however, automatically extend to their provision on the traditional single-employer DB model. The case for this model must rest on the presumption that we should not expect individual savers to bear and manage the risks of retirement income through market mechanisms. But this presupposition, at least when stated so starkly, is harder to support than many suppose, especially with regard to the private pension system. Granted, risks mean there will be losers--and in the matter of pensions, being a loser might mean poverty in old age. A society that protects people from many other risks inherent in a market economy will not likely accept pension arrangements that permit such an outcome. These considerations, however, do not so much support widespread private DB pensions as they do a social safety net for older Canadians. That is what the OAS-GIS system is for. Further, if the modesty of the support this safety net provides makes it desirable to top up the retirement incomes to some higher but still guaranteed level, that is what the CPP and QPP do.

Seen in the context of Canadian pension policy as a whole, then, the first answer to the observation that savers should not fall into poverty because of bad luck in the markets is that they already get insurance against the worst consequences of this eventuality. A more nuanced response, however, would note that, between the public system's minimum guarantees and the often deceptive promises of the classic DB model, there is a large space that other, and perhaps quite innovative, occupational pension arrangements might occupy. (20)

Potential Advantages of the Defined-Benefit Model

The most persuasive element in the case for the DB model is the observation that most individuals, being averse to financial risk and lacking the skills, time and resources to continually monitor and cope with it, gain from delegating the management of their retirement savings to specialists. An employer-sponsored DB plan permits such delegation. Certainly, it eliminates what many would see as an unfair situation in which two otherwise similar individuals, who have contributed identical amounts to their pension fund over their lifetimes, end up with different incomes because they activate their pensions on different dates, in different investment environments. Perhaps more significantly, by placing investment decisions in expert hands a DB plan has the potential to improve their quality, creating higher returns for savers and increasing the overall efficiency of the economy.

These considerations certainly argue for policies that put DB contracts at no legal or regulatory disadvantage to other retirement-saving arrangements. But do they support policies that would actively favour them? This distinction matters, because the advantages of delegating the management of retirement savings to experts through DB contracts may be less than they look in the abstract. We have already seen that during the 1990s in Canada and elsewhere, the allegedly superior risk-bearing and management capabilities of many DB plan sponsors led them to overestimate the premiums available from equity investment and to underestimate the hazards of asset-liability mismatches, especially in a world transitioning to low and stable inflation.

To put it another way, the undoubted ability of DB plans to relieve their members of the responsibility for managing risk does not extend to the more important matter of ensuring that they also cease to bear it. Policy should neither assume that it does, nor mislead the public into making this same assumption.

Limits on the Reliability of Defined-Benefit Guarantees

In the private sector, sponsors whose businesses are well protected from the competitive shocks of the marketplace seem best able credibly to promise defined-benefit pensions. Whether such protection comes from regulation or simply from market power, however, it can disappear. The airline industry, for example, has seen many widely publicized pension plan problems since its deregulation, and North American car and steel makers have experienced similar pension plan troubles in the face of foreign competition.

In the public sector, the DB model appears more viable, because plan sponsors can force taxpayers to meet their commitments should their pension fund managers let them down. Yet that same fact casts doubt on the wisdom of the model for this sector too. Ability to shift costs onto third parties does not promote responsible behaviour from either the sponsors or the beneficiaries of public-sector pension plans. In the event of trouble, future taxpayers are all too likely to end up the losers. And while defaults when taxpayers become unable or unwilling to fund the commitments governments have made on their behalf have been rare in Canada, they have occurred regularly throughout history in many places in the world. In view of the pressure demographic change alone will put on government programs in the decades ahead, at least partial defaults should never be dismissed from the realm of the possible.

Unforeseen economic changes, in short, can make even the most secure private-sector sponsors vulnerable to displacement, and can also affect government's ability to service their obligations. In both sectors, then, the capacity of sponsors to backstop the managers of their pension funds when things go wrong has limits. These uncertainties make flexibility and responsiveness to changing circumstances a desirable feature of any pension regime. Some recently suggested measures to help shore up DB plans, however effective in the short term, look unattractive from this broader perspective.

Extending government-sponsored insurance for DB plans beyond the limited regime that now exists in the single province of Ontario, for example, has so far been resisted, and rightly so (Robson 2005). If membership in such programs is voluntary, they will surfer from adverse selection: only bad risks will opt in. If membership is compulsory, such programs amount to little more than devices for bailing out badly run plans, first at the expense of plans that are more prudently managed and more reasonably negotiated and then, should matters deteriorate to the point of putting the public insurer under stress, at the expense of the taxpayer. (21)

Policies that encouraged or even mandated better matching of cashflows from investment and payment obligations would also, at least in the short run, create other problems. (22) Pension funds which sought close matching between their assets and liabilities would want to hold assets whose returns were strongly correlated with, and perhaps indexed to, consumer prices or even money wages. At present, we think that only governments, and perhaps only the federal government in particular, could issue such securities on a scale sufficient to satisfy widespread demand for them. But a non-governmental pension sector invested almost entirely in government securities would have little capacity to finance growth and innovation in the private sector. Indeed, because the taxpayer is responsible for servicing public debt, such a development would turn private plans into intermediaries in what would amount to a new type of pay-as-you-go public scheme. Until the day comes when more real-return assets such as infrastructure are available to investors in markets as liquid as those for shares and bonds, the immediate impact of mandated asset-liability matching would likely be to raise the projected costs of DB plans to an extent that would accelerate their disappearance.

The forces that undermine the single-employer DB plan's ability to "guarantee" benefits do not end there. The accrual of benefits in a typical DB plan over a worker's career is not neutral as regards place of work or time of retirement. Benefits do not vest at all at the outset of a worker's time with a single employer, and the full value of the package one employer offers will not typically travel with an employee, or be compensated by the package offered by a subsequent employer, when a person switches jobs. So DB plans may force employees to stay in an inferior job or accept a cut in pension. The uneven time-path of accruals tends to tip employees into retirement at a given age. In the face of imperfectly anticipated increases in longevity, this feature--which has also tended to lock plans into a higher ratio of pensioners to contributors than they expected--has undermined the ability of private-sector sponsors to deliver the benefits anticipated by their plans. It has not yet undermined the ability of public-sector sponsors to deliver anticipated benefits, but government employees retire years earlier than their private-sector counterparts, and the prospect that taxpayers will indefinitely underwrite promises that government-employee pension plans cannot support seems unlikely. (23) In short, the classic DB model is not necessarily attractive as a basis for pension policy in a country that wishes to encourage private saving, and more generally to promote efficient allocation of capital and labour to maintain and raise living standards in the 21st century.

The Alternatives

A key implication of the discussion to this point is that the classic DB plan may, in fact, impose costs on people who are not party to the contract. If that is so, the presumption that policy should favour such plans looks weak; it may, in fact, be better for policy to encourage a transition away from them. That suggestion brings into sharp focus another set of problems. Money-purchase pension plans are undoubtedly better at confronting their participants with risk and at providing benefits that are in line with what their assets can support, but currently available money-purchase plans have a number of important drawbacks.

Registered Retirement Savings and Defined-Contribution Plans

Some problems affecting money-purchase pensions in Canada, whether through defined-contribution (DC) pension plans or RRSPs, are easily resolved. In particular, limits on contributions to these plans make no more sense than their counterparts on DB plans; amendments to the Income Tax Act should raise or abolish them. But others, particularly those associated with the way in which risks are currently managed under such arrangements, are harder to fix.

Consider, for example, the investment risks an RRSP owner runs. Like a sponsor of a DB plan, this individual can invest in assets that hedge him or her against risks. The market risk involved in having to buy an annuity on the day of retirement is reduced by the options of transferring retirement savings into a Life Income Fund (LIF), in the case of group RSP assets, or a Retirement Income Fund (RIF) in the case of an individual's RRSP assets. Inflation-indexed annuities are expensive, but they are available. The option to hold deferred annuities within a DC plan enables a saver to introduce what amounts to a self-created DB element into it by locking in a given money pension long before retirement. But none of these is straightforward for a nonexpert who gets little chance to gain experience before making critical life-path decisions.

A closely related issue is administrative costs. Although DC plans shift responsibility for risk management from plan sponsors to members, to be viable and attractive, they need to devote resources to advising members on how to cope with this task. The sponsor's administrative burden is distributed differently among tasks under the DC model, then, rather than being obviously reduced outright. As a result, particularly for sponsors of small plans, it may not provide quite as much protection from the hazards and costs associated with the DB model as it might initially seem to promise.

Individual and employer-sponsored RRSPs feature high investment-management costs as well. Competition should lower these fees over time, and the lower charges typical in the United States suggest that this will happen in Canada. Even so, investing in exchange-traded funds for the sake of lower management fees will not change the fact that, once the time and effort that plan holders and their financial advisers devote to portfolio management decisions is taken into account, a myriad individual RRSPs will cost more in aggregate to administer than would a few pooled plans. Nor does it alter the fact that pooled plans would be able to offer a greater degree of partial insurance of benefits than small plans, should that be what savers desire.

Reduced ability to participate in certain type of investments is also a drawback of RRSPs and small money-purchase plans. The popularity of income trusts arose largely because ordinary dividend-paying shares are unattractive investments for retirement saving plans. The reason: the dividend tax credit which compensates investors outside such plans for the taxation of income at the corporate level is not available to investors in tax-deferred accounts. Ownership of income-producing assets outside the corporate structure is harder for smaller investors. So is participation in infrastructure and other less liquid assets that are well suited to retirement saving.

Perhaps, then, it is time for Canadians to consider some alternative models that would avoid some of the agency problems of traditional DB plans, while offering some degree of assurance about benefits. They would also avoid some of the administrative complexities of currently existing money-purchase plans, while retaining the greater transparency these provide about the risks inherent in market-based pensions. Some commentators (for example Ambachtsheer 2006) have presented ideas about how such plans might work. Although space does not permit us to elaborate at length here--and many potential features need more investigation and debate--some desirable characteristics of such plans are clear from the discussion so far.

Such plans would need to be predominantly money-purchase, with any minimum guaranteed benefit being small enough in proportion to be securable at reasonable cost. They would need to pool investment risk across a large number of individuals at reasonable administrative cost. They could impose contribution rates--perhaps mandatory, perhaps a default rate with participants accepting the consequences of deviating from it--geared, in light of evolving investment experience, to providing a target payout. They could impose a portfolio--again, perhaps a default portfolio, with the participants accepting the consequences of deviating from it--that would ensure that individuals were increasingly insulated against movements in annuity prices as they approached retirement.

In the United States, TIAA-CREF (Teachers Insurance and Annuity Association-College Retirement Equities Fund) is a prime example of a successful not-for-profit, arm's-length provider of defined-contribution pensions (among other services) to a particular group (academics, research, health and cultural workers). We know that some existing money-purchase plans in Canada have been approached by outside employers wanting to join them. And there already exist some "mixed" plans that provide pensions that depend upon market returns, subject to a minimum payment, and with provision for adjustments in contributions if necessary.

A key challenge for Canadians concerned about the future of the contractual/voluntary part of their pension system is to define more precisely the requirements of a healthy collective contributory system along the lines just described, and focus on the public policy actions and private entrepreneurship necessary to bring it into existence. The problems of the classic single-employer DB plan need addressing, but attempts to re-establish a "golden age" of DB plans that never truly existed would be a dangerous distraction from the urgent task of outlining a more promising alternative.

Conclusion

The problems facing employer-sponsored pensions in Canada are real and pressing. They call for a two-pronged solution.

First, many existing DB plans are in deficit, and their members' pensions are at risk. We salute many of the steps taken thus far to alleviate this situation: in particular, some flexibility in the timing of, and tools for, shoring up balance sheets. More is needed: tax provisions and federal and provincial laws that encourage underfunded plans must change. Beyond this, however, any search for further fixes in relaxed accounting standards, unwarranted assurance that some plans--especially in the public sector--are not at risk, or wider government provision of pension insurance are not appropriate. They are likely to make matters worse rather than better.

Second, Canadians must understand that current problems in the DB sector are not simply the result of a series of unfortunate financial events. They are symptomatic of deeper problems in the single-employer DB model. It follows, then, that the longer-run challenge is not to prop up and even expand plans built on this model. While the money-purchase alternatives that currently exist are far from perfect, this does not justify reflexive favouring of a DB model that has, after all, failed to deliver on its promises. The fact that there are flaws in both models argues first for policies that are neutral between them, permitting those for whom these flaws do not constitute prohibitive barriers to continue to choose them on their respective merits. But second, and more important, policy needs to encourage a search for new models that can offer Canadians more attractive ways to pool their resources and save for retirement in pension plans that mitigate not only financial risk and longevity risk, but agency risk as well.

We gratefully acknowledge helpful comments from Robin Banerjee, David Bond, Peter Drake, Yvan Guillemette, Gerald Goldstein, Claude Lamoureux, Michael Landry, Nick Le Pan, Finn Poschmann and several anonymous readers of earlier drafts. We accept sole responsibility for the contents of the final version, errors and omissions included.

References

Ambachtsheer, Keith. 2004. "Cleaning Up The Pensions Mess: Why It Will Take More Than Money." Backgrounder 78. Toronto. C. D. Howe Institute.

--. 2006. "Building Better Pensions on a 'fair value' Foundation." In Niels Kortleve, Theo Nyman and Eduard Ponds (eds.) Fair Value and Pension Fund Management. Amsterdam: Elsevier.

Anderson, Robert. 2005. "Trusteed Pension Plans and Funds, 1990 to 2004." In Canada's Retirement Income Programs. Ottawa: Statistics Canada.

--. 2006. "An Update on the Funding of Defined Benefit Pension Plans in Canada." Bank of Canada Financial System Review, (December): 43-50.

Armstrong, Jim and Jack Selody. 2005. "Strengthening Defined Benefit Pension Plans." Bank of Canada Financial System Review, December.

Association of Canadian Pension Management. 2005. Back from the Brink: Securing the Future of Defined Benefit Pension Plans. Toronto.

Corden, Sandra. 2006. "Private Pension Outlook Worsens, Regulator Warns." Globe and Mail Report on Business, Monday, March 27, p. B2.

Dimson, Elroy, Paul Marsh and Mike Staunton. 2006. "The Worldwide Equity Premium: a Smaller Puzzle." London, UK, London Business School. Mimeo.

Dodge, David. 2005/06. "Economic and Financial Efficiency: the Importance of Pension Plans." Remarks to l'Association des MBA du Quebec (AMBAQ), Montreal Quebec, November 9, 2005, Bank of Canada Review, (Winter): 55-59.

--. 2007. "A Sound Pension System: Handling Risk Appropriately," Remarks to the Conference Board of Canada 2007 Pension Summit Toronto, Ontario, May 10.

Fernandez, Pablo. 2006. "Equity Premium: Historical, Expected, Required and Implied." IESE Business School International Center for Financial Research Working Paper 661. University of Navarra. December.

Financial Services Commission of Ontario. 2005. Funding Defined Benefit Pension Plans: Risk Based Supervision in Ontario. Overview and Selected Findings 2000-2004. September.

Horner, Keith. 2007. "Emerging Challenges and Policy Responses within Canada's Retirement Income System: a Focus on Workplace Pensions, Private Savings and Income Security." Working draft prepared for Retirement and Aging Division, Strategic Policy Branch--Children and Families, Human Resources Canada (mimeo).

Kesselman, Jonathan and Finn Poschmann. 2001. New Options for Retirement Savings: Tax-Prepaid Savings Plans. C.D. Howe Institute Commentary 149.

Kocherlakota, Narayana. 1994. "The Equity Premium: It's Still a Puzzle." Journal of Economic Literature 34. March.

Maser, Karen and Robert Anderson. 2005. "An Introduction to Canada's Retirement Income Programs." In Canada's Retirement Income Programs. Ottawa, Statistics Canada.

Mehra, Rajnish and Edward Prescott. 1985. "The Equity Premium: A Puzzle." Journal of Monetary Economics 15: 145-161.

Office of the Chief Actuary. 2006. Actuarial Report on the Pension Plan for the Public Service of Canada as at 31 March 2005. Ottawa: Office of the Superintendent of Financial Institutions.

Office of the Superintendent of Financial Institutions (OSFI). 2005. Annual Report 2004-2005, Ottawa.

Ontario. 2007. Expert Commission on Pensions: Terms of Reference. <www.pensionreview.on.ca/english/about.html> (accessed March 13, 2007).

Robson, William. 2005. "Off the Hook, For Now: Taxpayers Should Hope that Talk of Federal Pension Guarantees Ends With the Minority Government." C.D. Howe Institute e-brief. November.

--. 2006. "Time and Money: Tracking the Fiscal Impact of Demographic Change in Canada." C.D. Howe Institute e-brief. October.

--. 2007. "Found Money: Matching Canadians' Savings with their Infrastructure Needs." C.D. Howe Institute e-brief. March.

Statistics Canada. 2003. Canada's Retirement Income Programs: A Statistical Overview (1990-2000). Ottawa.

Schembari, Patricia. 2004. "Employer-Sponsored Pension Plans over the Last 30 Years." In Canada's Retirement Income Programs. Ottawa. Statistics Canada.

Tuer, Eric and Elizabeth Woodman. 2005. "Recent trends in Canadian Defined-Benefit Pension Sector Investment and Risk Management." Bank of Canada Review (Summer): 21-35.

(1) The mandate of the Expert Commission on Pensions recently established by the Ontario government, for example, gives pride of place to "maintaining and encouraging the system of defined benefit pension plans" in the list of principles to guide the Commission's work (Ontario 2007, 1). Bank of Canada Governor David Dodge has also recently mounted a strong defence of the model. See Dodge (2007).

(2) One step toward addressing this problem is straightforward: letting people save out of post-tax income and exempting the proceeds of that saving from taxation (as proposed by Kesselman and Poschmann 2001) would mitigate it. The changing age structure of Canada's population makes a more thorough overhaul of the income and asset tests that create these disincentives appropriate, but that issue lies beyond the range of this Commentary.

(3) Maser and Armstrong (2005, Table 1), and Horner (2007, 36-37). Horner also reports that the incidence of pension saving in the 45-64 age group rises sharply from around 30 percent at an income of $10,000 to over 90 percent at $60,000 and remains more or less stable at higher income levels. RPP membership does, however, fall back above this income level from almost 70 percent to just over 50 percent at $130,000.

(4) This danger depends in part on how closely investors and their agents monitor individual pension fund balance sheets. Since pension regulators in Canada normally assess individual funds only once every three years, the information needed for timely monitoring may not be easily available when markets are volatile, as they were at the turn of the millennium. However, the latest study to refer to this matter (Armstrong 2006) concludes that "the direct consequences for the Canadian financial system of current pension deficits are not large" (p.50).

(5) Statistics Canada (2003, Tables 3 A5 and 3 A7) shows the fraction of paid workers covered by DB plans shrinking faster in the 1990s--from 41 percent in 1991 to 33 in 2000--than the fraction covered by plans of all types--from 45 to 41 percent, a picture reinforced by more detailed data from Horner (2007, Charts 3.5-3.8).

(6) The main exclusions from the FSCO sample are seven large public-sector plans with more than a million members, pensioners and other beneficiaries, and 776 "designated plans" (retirement compensation accounts) which mainly cover "connected persons and/or highly paid executives." Patricia Schembari (2004, Figure 1, 7-9), reports, as of January 1st, 2004, 7,014 DB plans, 7,507 defined contribution plans, and a relatively small number (fewer than 2,000) of "mixed" plans in Canada, with membership of 4.56 million, 0.88 million and 0.15 million respectively. The FSCO sample thus covers more than a quarter of total plan membership in Canada.

(7) About 10 percent of all private sector pensions plans fall under federal supervision (Armstrong and Selody 2005, 29-36); information on them presented here is drawn from OSFI (2004/05, 45 et seq.), and Corden (2006).

(8) Mercer Human Resources Consulting, "Report on the Funding Status of Canadian Registered Pension Plans and Five Year Financial Projections," unpublished, August 15, 2006. The ratio of assets in insolvent plans to total assets in the Mercer sample fell from 79 percent to 44 percent between the end of 2003 and May 2006. The sample studied included 847 plans for 2003, and 761 for 2006.

(9) www.watsonwyatt.com/canada-english/services/retirement/ret2.asp, visited May 1, 2007.

(10) This argument supports using something akin to the commuted value standard now in use, although there is room for debate about the exact formula.

(11) Narayana Kocherlakota (1994) provides a still useful survey of some of this literature, paying particular attention to the difficulty of interpreting observed equity premiums as compensation for risk. Fernandez (2006) emphasizes the distinction among historical, expected and implied equity premiums. Dimson, Marsh and Staunton (2006) use data from 17 national equity markets from 1900 to 2005 to calculate a premium in each that averages about 4 percentage points annually. Fernandez (2006) notes, however, that the problems of bridging periods when equity markets were closed and omitting periods (such as the early 1920s in Germany) when equity investors were wiped out, may mean that these estimates are too high.

(12) Ambachtsheer (2004, 1) puts recent estimates of the equity premium at "a very modest +1 percent to 2 percent, with a wide variation of possible results around this expectation for the next 10 years, or even longer." Dimson, Marsh and Staunton (2006) suggest that international investors might now anticipate a premium of equity over bond returns in the 3-to-3.5 percent range. Surveys of academics and financial professionals in the late 1990s and early 2000s yielded average expected premiums between 3 and 7 percent, with a tendency toward lower numbers in more recent surveys (Fernandez 2006, 13-14).

(13) These figures are for all pension funds, not just those associated with DB plans. The proportion of plan assets in stocks grew from 29 percent of portfolios in 1990 to 38 percent in 1997, but fell back to 30 percent in 2000. Anderson (p. 3) notes that funds' investment "strategy changed to one of greater diversification to lessen risk in 1997. The main diversification was to pooled investment funds, and to a lesser extent, real estate." The diversification in question was thus within the class of real assets, and seems to have involved no move towards a heavier weighting for nominal securities.

(14) Ambachtsheer (2004, 3) puts the total deterioration of pension plan balance sheets from 2000 to 2003 at $180 billion, almost one-third of their total value.

(15) See, for two expressions of this opinion among many others, Association of Canadian Pension Management (2005), and Dodge (2005/06). A useful summary of these asymmetries is "Box: key regulatory influences on DB pension funding," in Tuer and Woodman (2005).

(16) For a brief discussion of the case law underlying this problem, and the argument for addressing it by legislation, see Association of Canadian Pension Management (2005, 9-11).

(17) Over the year to the time of writing, the difference in yield between a nominal 30-year bond and its real-return equivalent was 2.5 percentage points--considerably more than the expected inflation rate in any survey of which we are aware.

(18) More securitization of infrastructure investments, for example, would be helpful (Robson 2007).

(19) Ambachtsheer (2004, 2006) has been a forceful exponent of this view. He argues (2006, 8) that these failings arise from the origin and evolution of workplace pension plans in North America as voluntary schemes, usually initiated by employers who initially took the view "that pensions are more perks than formal components of total compensation." He suggests that, had these plans arisen out of legislative requirements, their mandatory nature would have forced government and sponsors to collaborate on a framework to ensure that they were better managed from the outset.

(20) Disinterested public policymakers might encourage the spread of newer occupational plans, lest increasing resort to the simple DC/RRSP model, with all its drawbacks, begin to make a significant expansion of public pension schemes politically attractive. Policymakers more concerned with their own pension arrangements might also consider the greater controversy that would attach to the continued use of the classic DB model in the public sector should these arrangements disappear in the private sector and no satisfactory substitute emerge.

(21) The U.S. Pension Guarantee Corporation is already in difficulties, having had to take over the.... liabilities failed private sector schemes--most recently of United Airlines--and presents policymakers with the unpalatable choice between letting it rail and bailing it out at the taxpayers' expense. The Ontario government's Pension Benefit Guarantee Fund is poorly financed and would need tax support to hand|e even a moderate-size failure.

(22) Recent amendments to pension legislation in Quebec permit regulation to impose provisions for adverse deviations on plans, depending on their degree of mismatch. At the time of writing, no such regulations have been promulgated, and we hesitate to endorse a regulatory approach over market disciplines.

(23) The most recent actuarial report on the federal public-service pension plan showed that, discounted at an ultimate real rate of return of 1.73 percent--roughly the then current yield on the federal government's 30-year real-return bond--the normal cost to maintain full funding was almost 33 percent of pay, an astonishingly expensive promise (OCA 2006, 59).

David Laidler is a Fellow-in-Residence at the C.D. Howe Institute and Professor Emeritus at the University of Western Ontario.

William B. P. Robson is President and Chief Executive Officer of the C.D. Howe Institute.