Many companies that still have defined benefit pension schemes are faced with the risks of a pension fund’s solvency levels dwindling through poor asset performance. It's for that reason that many trustees have found themselves considering a structured product that matches cash flow requirement of the fund over the next 15 years or so, eliminating that risk. My concern is that many trustees are underestimating the potential risks of these simple solutions.
The cost of simplicity
The “simple solution” is to buy an instrument from a bank that provides the exact cash flows predicted above from that bank. That will cost, say, 70% of the value of the fund; the balance can then be invested in more risky assets, because the important risks are eliminated and over the long term the risky portfolio will outperform. This structure is especially appealing if the fund, on a fairly low discount rate basis, is either near fully funded or in surplus. What’s more, it’s argued all costs are included in the structure and so the trustees have no other costs.
Some of the problems with these solutions include:
* Model risk, or the potential error in pensioner payment forecasting. That risk is a problem that unfortunately affects all matching solutions and is certainly not limited to only structured products. A mismatch means you’ll either need more cash if pensioners live longer or have to reinvest (at uncertain levels) if pensioners have a higher initial mortality. More importantly, illiquidity problems are exacerbated when using structured products, given their limited flexibility.
* Problems of duration versus a proportional matching strategy. This is the more serious problem. A duration match, is when the matching strategy matches the first, say, 10 or 15 years of cash flow. A proportional match, is one where a certain percentage of all cash flows are matched: eg, 70% of cash flows over the next 80 years.
If a fund starts today with a fully-funded solvency position and, over time, asset prices and liability valuations vary, growth to pensioners will be declared depending on the success of the strategy. At any stage you can value the dynamic assets and liabilities and compare to determine a funding level. The graph shows the probability of that future funding level in three years’ time being between the levels indicated: ie, 0%-50%, 50%-75%, 75%-95%, etc, based on the different strategies. (See graph.)
The reason I don’t like the duration match is simple. As seen, while the proportional match almost eliminates the possibility of a deficit arising greater than 25% the duration match hasn’t eliminated that risk.
Yes, that sounds incredible. But the problem lies in “compounding”, and is a great example of when your best friend becomes your enemy life can become rather tough. By illustration, if inflation-linked bond yields drop substantially, accompanied by underperformance or even losses on the risky (equity) portfolio (unmatched component), when you measure the liability increase a fall in solvency becomes a possibility. The reason being the cash flows left unmatched have the longest duration (ie, cash flows after the first 15 years, where compounding of mismatch error has the greatest impact) and those increase substantially in value, while in the market scenario described the risky portfolio would have underperformed that substantially. That will result in severe underperformance of the duration match versus the proportional match, where longer-term cash flows will have a lower impact.
So compounding doesn’t favour duration matches.
So why is this 15-year cash flow solution sold? Because 15 years is all the banks are willing to guarantee to pension funds, for a variety of reasons.
Credit, liquidity and regulatory risks
First, it’s clear if you buy a product or note from a bank you’re exposed to the credit risk of that bank. That risk is potentially disastrous, as those trustees who bought matching Lehman structures of the same kind now understand.
Generally, SA’s bigger banks issue paper at credit spreads of between 0,5% and 4% over risk-free, depending on the ranking of the instrument. It’s therefore interesting to contrast the tiny spreads offered by banks on similar risk-structured products that leave pensioners way under rewarded compared to the norm for the same bank credit risk they’d take on if they bought that same bank’s bonds. By buying those instruments directly trustees would be able to increase pension growth to the same extent as the spread they would have foregone without taking on any extra credit risk.
But while that would sort out the pricing it wouldn’t eliminate the credit risk. So in order to alleviate the problem, structured product teams provide collateral portfolios of “Government guaranteed” bonds as security for the banks’ credit risk. However, those bonds are often highly illiquid. In that case, where the bank defaults – despite the collateral having a Government guarantee – liquidation of the illiquid Government bond may have pricing move against the fund over and above any legal costs involved.
However, the primary liquidity risk lies not in the collateral but in the liquidation of the cash flow note. If the fund wants to liquidate, there’s no open market to unwind the structure. Costs to pensioners of unwinding all or a part of the structure can be in excess of 5%.
So if mortality is less favourable for the finances of the fund – ie, pensioners live longer – or regulations change resulting in the need to liquidate the positions there could be great expense in resolving the problem.
Regulation 28 limits the exposure in any fund to the credit risk of any one bank. PF 133 goes further, in that all derivatives must be looked at on a look-through basis to the underlying risk. In theory, that actually means all banking structures of this type fall foul of the regulation, despite the collateral and margining techniques used to keep exposure and market value of the structures within guidelines.
Opportunity costs
The final risk is to pensioners, in the form of an opportunity cost also related to the inflexibility of structured products. The returns of that “note” or structured product can’t be enhanced through the use of alternative yet appropriate instruments providing attractive credit spreads (as discussed earlier) or the switching of credits through time, or repos and carries.
So what’s the better solution?
Given the problems mentioned, plus the fact the balance of the portfolio (unmatched component) takes on far higher levels of risk in the portfolio, there is a better way. The answer is clear and it lies in the same principles and strategies taken on by the bank to manage its risk in selling those structures to pension funds.
These principles recognise that:
* Risk is compound and can’t be bucketed simplistically.
* But all risk must be rewarded fairly.
* Everything is priced off a (credit-adjusted) yield curve and other market variables, so there are no free lunches.
* Elimination of risk is expensive and may not be certain.
* If a structure is sold as having no cost, the reality is probably far from that (as there are potentially hidden costs).
* Holding on to flexibility of capital is important.
* Simplicity is expensive and often misguided.
Certainly, in Britain that’s acknowledged and those are the strategies being implemented after considerable school fees being paid by their trustees. I can only hope SA’s trustees will choose to not learn through the same mistakes their peers have already made.
JARRED GLANSBEEK
RisCura
GLANSBEEK is CEO at RisCura. Previously an executive director at RMB Asset Management he has substantial experience in quantitative portfolio management techniques and complex instrument pricing and design, as well as judgmental and active asset management. Glansbeek pioneered and implemented Liability Driven investing as early as the mid-Nineties while being part of a team managing structured products and pension fund solutions. After leaving the FirstRand Group he started RisCura, currently one of SA’s largest asset and liability consultants. It’s also regarded as the industry’s leading risk manager.