I am often asked the question: “Are you doing any liability-driven investing (LDI)?” My response is typically something like: “yes, the only reason we have assets is to pay liabilities. We are very conscious of the risks in our liability payment stream and we are constantly looking for ways to cost effectively hedge those risks.” The person asking the question will then ask something like: “How do you lengthen your fixed income duration, with long bonds, derivatives or leverage.” My response: “none of the above, we are short duration right now, between 4.5 and 5 years on our total fixed income portfolio.”
This produces a puzzled look and I may start a long explanation of the apparent contradiction between ‘doing LDI’ and having a relatively short duration position in our bond portfolios.
It is unfortunate, but common, that terminology and ideas in our industry get taken over by service providers and attached to specific strategies or specific products.
LDI should be a broad concept where the strategies employed are very specific to each investor based on their liabilities, their own risk assessment and the cost of hedging those risks. What has happened, however, is that the acronym LDI has become associated with specific strategies or products aimed at one specific risk: interest rate risk. This approach runs the risk of hedging something that might not be a large risk and missing other risks that may in fact be much more material.
When thinking about LDI, I would suggest that plan sponsors and boards should start the process by looking at the risks first and the hedging strategy or products last. The questions, in order might be: What are the risks? How will the risks manifest themselves? Can the risks be quantified?
At WCB – Alberta the key risks to the liabilities are interest rate risk and inflation risk. Defined benefit (DB) pension plans would add longevity risk and maybe other risks depending on any unique plan or business characteristics.
These risks have been quantified at WCB – Alberta. The expected liability payments have a duration of approximately 10 years and each 25 basis point change in the discount rate has an impact of $140 million. The impact of inflation is seen directly on the cost of actual benefit payments as they are adjusted with a COLA on wage replacement benefits and are directly seen in health care payments.
In 2009 and 2010, with low inflation for both COLA and health care, there was an actuarial gain of $123 million – about $50 million per 1% per annum. This was an impact on cash payments for that year, but if the inflation assumption in the liability valuation is changed, then the impact is much larger; since it impacts the expected cash outflows for the entire future cash flow streams. This would have a similar impact as a change in the discount rate, 25 basis point = $140 million.
How will these risks manifest themselves?
The key issue is that interest risk impacts the value of liabilities by changing the assumed rate of return for investments or discount rate. I would call this a measurement risk as it does not actually change the actual cash outflows that we are trying to fund. It only changes the present value of the cash flows or the measurement of how much we think we need to have set aside today to pay future cash flows.
Inflation can have this type of impact as well, if we experience a period of high inflation, the actuaries may change the inflation assumption and if they change the inflation assumption the expected cash flows will be higher. If the real discount rate is unchanged then this will be offset, but that might not be the assumption used.
More important is that a period of high inflation will actually change the cash flows that go out the door. Wage replacement, health care and operating costs (or pension payments for DB plans) will be higher than expected. As opposed to a ‘measurement risk’ this is a real cash flow risk – real money out the door!
In my next post I will look more closely at that key piece risk puzzle – inflation risk.