Part three in our LDI series: LDI in action
IN PRINT ARCHIVE CIR Winter 2007
Part three in our LDI series
By Josephine E. Marks, F.C.i.a., C.F.a. Marks 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.
Portfolio management is an art as much as it is a science. The same holds true for liability-driven investing (LDI) or asset liability management. In a science, reproducing the same events under the same conditions results in the same consistent outcome. However, as discussed in the two previous articles in this series, it is virtually impossible to construct a portfolio that will perform exactly as expected relative to the liabilities of a going concern pension plan under all market scenarios. The art is in constructing a portfolio that has a high expectation of securing the pension promise, albeit with no guarantees.
To appreciate the artistic licence involved in portfolio selection, consider the varied approaches used by some of Canada’s leading funds. The Ontario Teachers’ Pension Plan (OTPP) uses an asset-liability model to validate its investment strategy. The current portfolio includes 45% equity for long-term growth, 20% fixed income for absolute returns against a nominal benchmark and 35% inflationsensitive assets (real return bonds, real estate, infrastructure, commodities and timberland) as a hedge for real liabilities.
OMERS (the Ontario Municipal Employees Retirement System), which is also pursuing a strategy based on the long-term real return requirements relative to the plan obligations, has a current portfolio mix that is 60% equity, 20% fixed income and 20% alternatives such as real estate and infrastructure. However, the long-term focus is to transition the portfolio to a 35% allocation to alternative investments as a source of strong predictable returns and consistent cash flow against future liabilities.
Smaller plans, whose investment strategies may be more constrained due to resource limitations, are also pursuing liability driven investing. For example, the Colleges of Applied Arts and Technology (CAAT) plan, is restructuring its portfolio to classify assets as either liabilityhedging (nominal bonds, real return bonds, real estate and infrastructure) or return-enhancing (equities, active overlay strategies) with an actual mix of 40:60 between these two categories. Among corporate plans, sponsors such as Nortel Networks Limited have responded to the LDI challenges implicit in new accounting standards by shifting the asset mix to reduce the equity exposure. Nortel would then consider the use of interest rate swaps to reduce the duration mismatch in the fixed income portfolio and the use of portable alpha products to enhance performance.
Matching versus risky assets
All of these plan sponsors may be deemed to share a common trait—they consider assets as falling into one of two categories, matching or risky. The matching assets (e.g. real return bonds, income-producing real estate, nominal bonds, infrastructure) tend to behave in a manner similar to nominal or real liabilities, provide cash flows to meet liability obligations, or are compatible with accounting regulations. The risky assets (e.g. public and private equity, absolute return strategies) generate enhanced returns to provide for the pensions actually promised if these are more generous than those supported by the matching assets.
How do risks get managed under this portfolio construct? Matching assets, by definition, should reduce funding risk. Risky assets are selected in traditional assetonly space, where risks are managed through prudent asset selection and diversification. Risk budgeting may be used to determine the allocations to these assets. Risky assets may secure additional return in the short term, through trading strategies (arguably a zero sum game), or in the long term, through the creation of economic wealth.
Portfolio construction within this liability-driven investing context
leads to two primary risk-reducing objectives:
Plan sponsors need to assume these risks within the overall portfolio risk tolerance, while taking market conditions into account.
It may be interesting, in light of the previous articles in this series, to examine how this portfolio model applies to the insurance industry. When the insurance industry sought to match assets and liabilities in the 1980s by reducing interest rate risk, portfolios were shifted towards fixed income assets. However, the industry still needed to assume risk so as to earn the return required to support more aggressive product development and pricing in a competitive market.
Why did the insurance industry primarily pursue credit risk in this regard? First, it could easily be assumed in conjunction with the fixed income assets. Derivative markets had not developed to separate credit risk from interest rate risk. In fact the industry faced a conundrum—if they were to invest in long bonds to mitigate the interest rate risk, they would be assuming long-dated credit risk, which was not ideal. As a result, the industry chose to use private debt at the long end because contractual provisions were more favourable to the lender than in the public markets. Second, the growing investment management industry had poached equity management talent from the insurance industry, with more attractive compensation packets. The insurers found it easier to find talented fixed income managers without deviating from their more constrained compensation models, and could create a differentiator in this space. Finally, regulatory charges for the cost of capital were far higher for equity assets (despite the upside potential) than for fixed income assets, even those below investment grade. Thus, for the insurance industry, the same assets provided both the matching and the risky characteristics.
Looking at Figure 1, however, which portfolio is riskier? Portfolio A has a traditional 60:40 mix which is 40% matched and 60% leveraged to invest in equities, and Portfolio B is 100% matched to liabilities with a portfolio overlay of up to 60% in equity exposure. Portfolio A has borrowed long to invest in equities while portfolio B has borrowed short to invest in equities. Indeed, under most scenarios, for an upward-sloping yield curve, portfolio B has better risk/return characteristics than portfolio A.
A more pragmatic way to define leverage within the context of portfolio risk is that a leveraged portfolio assumes more than one risk per invested dollar, or has the potential to lose more than one dollar per dollar of net assets in a worst-case scenario. Derivative instruments allow such financial leverage to take place without actually borrowing funds. Financial leverage can be used in conjunction with the separation of alpha and beta to create more effective portfolios. For example, pension funds may source market returns (beta) using derivatives (such as total return equity swaps) without deploying capital. Or they could source absolute return strategies (alpha) in long/short equity selection, credit, currency, or arbitrage opportunities, also without deploying capital.
It may be argued that alpha strategies are a zero sum game and that pension funds would be better off focusing their efforts elsewhere. Successful stock selection (source of the common long/short alpha strategies) may be attributable to luck rather than skill unless one has many years of track record. However, inefficient markets such as credit or currency may also allow for outperformance due to skill. Arbitrage opportunities are usually fleeting, except for those that require government intervention to end (e.g. tax or capital arbitrage).
Before undertaking leveraged derivative activity, it is not just important for a plan sponsor to thoughtfully assess the risk and return potential, they should also ensure that the right governance, oversight, education, alignment of interests, personnel, culture and controls are in place. Along with its benefits, derivatives use involves higher risks than physical assets due to the complexity of the instruments.
The Leveraged Portfolio
Derivatives can be used to mitigate the downside risk, either at a net cost to the fund or by sacrificing some of the upside potential (e.g. a collar strategy). It is important to assess such alternatives in advance—what matters is not so much which of the many possible scenarios unfolds (the luck of the draw) but what the range of potential outcomes was a priori. Why is the a priori perspective important? Suppose there are two potential portfolios—the first is expected to deliver a real return of -3% to 5% relative to liabilities, whereas the second is expected to deliver a real return of 1% to 3% relative to liabilities. A risk averse, return-seeking investor might well select the latter a priori. But what if the first portfolio actually delivered a return of 4% while the second actually delivered 2%? In hindsight, the same investor might believe that the first portfolio had been the preferred choice. Portfolio strategy needs to be set in advance, within the risk tolerance established under the LDI framework, and not be subject to second-guessing after the fact.
Implementation of the LDI principles described above leads to a number of portfolio implications. In practice, any portfolio construction that purports to offer asset-liability matching for a going concern pension plan should be viewed with great skepticism. However, if plan sponsors pursue assets whose interest rate sensitivity or cash flow characteristics better mirror the liabilities, this will increase overall appetite for real return bonds, income-producing real estate, or income-generating infrastructure.
Furthermore, addressing the duration mismatch will reduce the support for managing fixed income by taking duration bets—a winning strategy during the last twenty years of declining interest rates, especially when the benchmarks had short durations. Taking short positions during periods of rising rates is more problematic, given the level of funding risk relative to liabilities should the investment view turn out to be wrong. Instead, fixed income managers will be expected to develop more diverse management styles involving strategies such as exploiting yield curve shifts and sector shifts, foreign debt (currency hedged), emerging market debt, private debt, structured credit, distressed debt and/or credit derivatives.
Building a more diversified risky portfolio may increase overall appetite for private equity, emerging markets, foreign real estate, and absolute return (hedge fund) strategies. While the former may involve long-term investing or creation of wealth, the latter may simply create what John Maynard Keynes referred to as “beauty pageant investing,” where market participants try to outperform the competition. In an efficient market, these strategies are zero sum games and should only be pursued if the manager has demonstrable skill or a competitive advantage (e.g. access to arbitrage opportunities that are not fully exploited).
Figure 3 and Table 1 illustrate the potential asset mix and return expectations for a traditional portfolio versus leveraged LDI portfolio. This structure assumes that a pure matching portfolio would be incapable of securing the pension promise, and that additional risk is required.
In the LDI portfolio, the core portfolio consists of physical assets while derivative overlays are used to enhance returns by assuming additional risk. Physical assets include matching fixed income assets and alternative assets for which derivative substitutes are not available. The alternative assets may be matching or risky assets. Derivative overlays may add public equity market risk (beta) or active management risk (alpha). In practice, it is unlikely that the LDI portfolio would be leveraged to this extent. Some of the equity market risk and some of the active management risk would come from the core portfolio of physical assets.
Asset class managers are usually benchmarked within each asset class, whether absolute or relative to market indices or competitors. Missing are true liability benchmarks that measure the progress of the total asset portfolio against the pension obligations, as opposed to the current practice of merely summing the asset class targets, to establish total fund targets.
Alignment of interests is problematic. Consultants or advisors do not have financial skin in the game, although they are motivated to maintain client satisfaction. Investment managers may have different levels of tolerance for portfolio risk based on the compensation structure. Some of these issues can be addressed by having an internal accountable team whose employment and core compensation depend on the overall success of the fund. Compensation should reward this team for providing a sound basis for portfolio construction, without encouraging an excessive focus on generating added value, which is a relatively small contributor to the success of a fund.
It is difficult to avoid conflicts of interest and address governance challenges. Hiring or selecting high-quality people who demonstrate integrity is a good first step. Similarly it is virtually impossible to create a portfolio structure that defeases the portfolio obligations. However, the consideration of LDI principles moves the industry in the right direction.