Lessons from the past
The strategies for pension fund investing have struggled with liabilities since the beginning.

By Josephine E. Marks, F.C.I.A., C.F.A. has had extensive industry experience in pensions, investments and asset-liability management, most recently as Chief Investment Officer for one of Canada’s largest public sector pension plans.

Pension fund sponsors are undergoing a shift in thinking. After years of presiding over contribution holidays and benefit improvements, plan sponsors and trustees are now grappling with unfunded positions and looking at their investment strategy through liability-driven investing glasses.

What is meant by liability-driven investing (LDI)? It’s akin to the know your client requirement of the retail investing world. Fundamentally, it refers to investing the assets with consideration given to the characteristics of the liabilities. At one extreme, this could mean setting up a portfolio that is perfectly matched to the cash flow of the liabilities. Alternatively, it could mean setting up a portfolio that is structured to have a very high probability of meeting all the liability obligations as market conditions unfold.

In this first instalment in a series of three on LDI, the historical context of pension fund investment strategies relative to liabilities is examined by reviewing the history of pension funding and of asset-liability management.

A new era
In the early days of many pension funds (1950s to 1960s), the funding vehicle was simple and secure. The plans were career average plans and funded using paid-up deferred annuity contracts purchased from an insurance company. For its part, the insurance company would invest the premiums received primarily in government bonds, monitoring returns and using model portfolios to ensure that the basis for pricing the annuities was sound. One might say that they were pursuing asset-driven pricing, or liability-driven investing in reverse.

Every year, the plan sponsor would purchase the accrued benefits at the guaranteed market rates with total peace of mind—after all, no Canadian insurance company had ever defaulted on its obligations and their balance sheets were prudently managed. Contributions were stable and affordable and the benefit design was competitive for the times. From time to time the insurance company would declare a dividend based on favourable experience relative to the conservative pricing assumptions and add to the benefits. It was quite reassuring for sponsors and members.

However, the structure was not very flexible and did not adapt well over time. With the onset of inflation in the 1970s and the evolution towards final-average plans, a more aggressive investment approach was developed. In theory, the plans could have continued to purchase paidup annuity contracts, but they would have had to purchase the accrued benefit plus the inflation and wage increase adjustment for all of the previously accrued benefits each year. It was much more appealing to invest the portfolio in a separate trust and allocate 60% to equities to provide some protection against inflation and to take advantage of the equity risk premium. Now the benefits could be enhanced as the actuary discounted the liabilities, assuming a three to five per cent equity risk premium in perpetuity on the portion of the fund invested in equities. A new era in pension funding had begun.

Fatal assumption?
In hindsight, it’s convenient to criticize the addition of equity risk to the mix as a fatal flaw in the history of pension funding. It introduced mismatch risk and risksharing between sponsors and members, with none of this being acknowledged or even disclosed at the time. However, there was a reasonable logic to the new approach under the conditions then prevalent. Equities were deemed to be as well correlated with inflation as any other asset class, bearing in mind the lack of issuance of real return bonds and lack of experience with alternative assets. The funding risks were deemed to be wholly owned by the plan sponsor, in that they expected to lay claim to plan surpluses or pay the amortization costs for experience deficits as they arose. Assumption of portfolio risk was a sensible business practice provided the risk was assumed in a prudent fashion—even the staid insurance companies were now pursuing more risk—adding more credit risk, real estate risk and illiquidity risk. The investment management industry evolved to convince plan sponsors that equity risk could be safely pursued using professional managers who would manage portfolios towards a stable long-term return while also delivering an active management value added premium of as much as 5 %.

Equity markets co-operated during this period as well, with Canadian equity returns averaging 12% per annum through the 1980s. As a result, the pension plan investment model remained relatively stable during this period. But events in the life insurance industry during the 1980s gave rise to some profound changes in that sector and the advent of asset-liability management (ALM).

Managing interest rate risk
One of the drivers influencing the evolution of ALM for the life insurance industry were the high profile failures of certain savings and loans companies in the U.S. Interest rates had become more volatile during this period, resulting in extreme interest rate mismatches on balance sheets. In practice, the term asset-liability management was often used synonymously with interest rate risk management.

The introduction of ALM in the insurance industry gave rise to formal asset segmentation, so the lines of business would have their own investment strategy and distinct portfolio. Each asset portfolio would be customized to the liabilities it backed, using factors such as duration, credit risk, liquidity and asset mix target and range. Interest-sensitive liabilities were typically backed by matching fixed income assets. But the insurance companies also assumed credit risk through corporate bonds, added illiquid private placements and mortgages in the mix, and included some income-producing real estate to the portfolio. This diversified portfolio provided enhanced returns while smoothing outcomes over a business cycle. Liabilities that were not interest-sensitivewere backed by more traditional portfolios, including some equities in the mix.

Likewise, pension plans also undertook assetliability studies in the 1990s. While most of these studies reconfirmed the 60/40 equity/bond mix, they also pointed out that growth in the liabilities had been outpacing the returns on the fixed income portfolios in a declining interest rate environment and that equities were a volatile asset class. Figure 1 shows the history of interest rates from 1936 to 2003 on long-term Government of Canada bonds (CANSIM 122487) as a reminder of how interest rates climbed through the 1970s after many years of comparative stability and then dramatically declined thereafter.

Figure 2 shows the history of the equity risk premium from 1975 to 2003, as represented by the difference between Canadian equity returns and the risk-free return during that same period. Although the average equity risk premium during that period was a respectable 4.2%, the annual experience was both variable and volatile, ranging from -31% to 32% on an annualized basis and from -12% to 23% on a three-year moving-average basis.

In response to these concerns, plan sponsors sought to better match assets and liabilities in the 1990s by adding long bonds and real return bonds to the portfolio and by diversifying away from public market equity risk—making more use of alternative assets such as real estate, private equity, infrastructure investments and hedge funds. Some of these asset classes provided modest levels of inflation protection, while others were selected simply due to their diversification benefits. The net effect would be to stabilize the funded ratio and to improve the risk-adjusted return expectations. Liabilitydriven investing re-emerged.

Search for stability
Despite the focus on asset-liability studies in the 1990s, few pension plans were prepared for the perfect storm that hit from 1999 to 2001. The market conditions, which combined an equity market downturn with declines in interest rates (thus further increasing liability values), wreaked havoc on the pension landscape.

Figure 3 demonstrates this graphically by showing the historical U.S. asset and liability returns from 1989 to 2005, as compiled by Ryan Labs. Asset returns represent those of a typical pension plan with a traditional 65/35 equity/bond asset mix while liability returns are those of a portfolio of nominal and real return bonds deemed to match the average liability stream. While asset returns easily outpaced liability returns through most of the 1990s, the funding gap that emerged from 1999 to 2001 caused the average funded ratio for U.S. plans to decline from 139% in 1999 to 73% in 2002, according to Ryan Labs.

Unfortunately, the market conditions associated with this period were not the only factors causing pressure on the pension system. Other drivers for pension review were:

• The ability of plans to build a buffer against poor market conditions was restricted by Revenue Canada regulations requiring plans to take contribution holidays or make benefit improvements when surplus exceeded 20% of liabilities or twice the normal cost.

• Development of new accounting regulations in other jurisdictions (the U.K. and the U.S.) which required that liabilities be valued according to current interest rates, thus further exacerbating mismatch risk when the assets weren’t interest-sensitive.

• Lack of symmetry between surplus and deficit ownership after the 2004 Supreme Court ruling on the Monsanto pension plan which granted surplus ownership on a partial wind-up to those employees affected, while leaving corporate sponsors accountable for pension deficits.

• Continuing conversions from defined benefit (DB) to defined contribution (DC) type plans, where plan sponsors transferred the investment and longevity risk to the members.

For corporate plan sponsors, analysis by Black1 and Tepper2 had demonstrated that higher returns on the pension fund did not necessarily enhance shareholder value. This was because the accounting treatment used for the pension plan gave rise to earnings volatility when the asset behaviour did not correspond with the changes in liability values. This volatility would be deemed by the investing market to increase the company’s financial risk. The company’s shareholders were not being compensated for bearing this additional risk. So corporate pension fund sponsors debated whether a more conservative investment approach might in fact enhance shareholder value.

For the public sector pension plans, the interest in liability-driven investing arose from a more direct objective—maintaining stability in benefits and contributions. Many of these plans had taken contribution holidays in the 1990s and now needed to restore contributions at the same time that the sponsors were under budgetary stress. Meanwhile, as the pension plans were maturing, the higher proportion of retired members meant that the funds also had less ability to spread the higher contribution rate over the active membership. From a benefit perspective, there was no appetite for eliminating any of the benefit improvements granted during the 1990s, but growing concern that unless the investment strategy fully supported the funding objective there might be pressure to reduce such benefits in future if market conditions were unfavourable again.

All of these factors combined to threaten the continued viability of the defined benefit plan model, and renew industry interest in liability-driven investing. Liability-driven investing and asset-liability management were certainly not new concepts, as has been noted earlier in this article. However, what was different this time were the risk management tools and portfolio construction capabilities to analyze and implement new portfolio strategies.

Evolution of managing risk
In 1952, Harry Markowitz defined portfolio risk as the standard deviation of portfolio returns and structured portfolios by creating asset mixes, which optimized such risk-adjusted returns.3 By 1993, extensive work had been done by Rom and Ferguson4 and others on understanding and analyzing downside or semi-variant risk, using the more intuitive definition of risk as being the risk of loss. This work had not been possible earlier due to the lack of computational capability to support the more advanced mathematics required. This approach became known as post-modern portfolio theory.

Unfortunately, the body of research work on portfolio construction had been developed using the principles of modern portfolio theory so very few pension practitioners explored the concepts associated with downside risk. Assessment of value-at-risk or traditional asset-liability studies assumed that risks were symmetrical and normally distributed. These shortcomings meant that much of the modelling work associated with asset-liability studies for pension plans failed to reflect the full scope and range of possible outcomes.

In addition to having access to better tools to assess the nature of risk (whether they were actively used or not), the 1990s saw many developments in terms of the market’s ability to decompose and recombine risks.

In managing the interest rate risk associated with their liabilities, insurance companies now used interest rate swaps and other fixed income derivatives. This enabled them to lengthen portfolio duration while not also extending the term of their credit risk. Later the use of credit derivatives would allow the converse —the assumption of credit risk disassociated from the interest rate risk.

Similarly, pension funds were able to use equity derivatives together with long/short equity management to separate market risk (beta) from the value added by active management (alpha). The introduction of hedge funds to the pension industry allowed for a wide range of uncorrelated alpha strategies to be implemented without assuming the associated market risk. These tools in portfolio construction allowed risks to be taken in a more precise fashion.

A framework for LDI
In developing a liability-driven investing framework, pension plan sponsors need to assess the funding strategy based on the following questions:
1. What funding risk could or should be assumed by the plan and who is ultimately bearing such risk?
2. What is the nature of the mismatch risk between assets and liabilities and how can this risk best be managed? Liability driven investing need not mean that pension funds eliminate risk from their portfolios, but that such risks should be acknowledged, understood and consistent with the stakeholders’ tolerance for risk. Understanding risk is one of the most important elements in designing an effective LDI strategy.

Acknowledgments
The author acknowledges the thought leadership of Malcolm Hamilton and John Por in originating some of the ideas expressed in this article.

References
1. Fischer Black, “The Tax Consequences of Long-Run Pension Policy,” Financial Analysts Journal, July – August 1980
2. Irwin Tepper, “Taxation and Corporate Pension Policy,” Journal of Finance, Vol 36, No 1, March 1981
3. Harry Markowitz, “Portfolio Selection,” Journal of Finance, Vol 7 No 1, March 1952
4. Brian Rom and Kathleen Ferguson, “Post-Modern Portfolio Theory Comes of Age,” Journal of Investing, Winter 1993

For a pdf version of this story, click here.