In the pension world, long-term expectations are always subject to short-term reassessment.
Ten to 20 years ago, a 7% long-term expected return on pension investments was considered very conservative. Now it is considered extremely aggressive. The long-term return assumption has steadily declined, from 7% to 6.5% to 6% to 5.75%…turning pension programs that were once thought to be affordable into unsustainable albatrosses.
For a while, actuaries resisted lowering the discount rates. After all, a long-term assumption is supposed to be long term.
But in the face of declining interest rates, this stand became untenable. When we determine an expected future return on our investments, the place to start is generally bond yields. This is the market’s return expectation for bonds. Hold bonds to maturity, and this is what you get. Given a current yield on universe bonds of around 2.3%, even a 6% return assumption becomes a huge stretch, and 7% looks completely out of reach.
If we assume a typical asset mix of 60% stocks and 40% bonds, the contribution to expected return on the bond portfolio would be 0.92%, so to get to 6% we would need 5.08% from the 60% that is in stocks. The return on the stock portfolio would therefore need to be close to 8.5%—not impossible but highly optimistic.
Even after we have decreased our return assumptions to all-time lows, they may still be overstated. There are two reasons for this:
- we may have entered an extended period of low returns, lower than what we have yet assumed; and
- we have not fully accounted for volatility.
There are many who believe that the financial crisis that started in 2007 represented a watershed in the economy. It commenced a period of deleveraging, in which, for a number of years, all of us—countries, corporations and individuals—will be under pressure to pay down debt to more manageable levels. The result will be lower economic growth, disinflationary pressures, sustained low interest rates and generally a lower-return investment environment.
But also, the methodology for determining long-term return assumptions has always been suspect in its lack of explicit regard for volatility.
When we develop an expected return using a building blocks method—such as 0.9% from the bond portion and 5.1% from the stock portion equals 6%—we are basically saying that a 6% return, in any given year, is a reasonable expectation.
But the range around this expectation is huge. The standard deviation of a typical balanced fund is roughly 10%. So, to use an extreme (two standard deviation) example, we could, on average, be achieving a 6% return by getting returns over the next couple of years of 26% and -14%. The arithmetic average is a 6% per year return, but the compounded annualized return is around 4.1%.
If a $100-million plan is fully funded with a 6% return assumption and it actually gets a 6% return each year for six years, at the end of the period it will still be fully funded, assuming all the non-investment assumptions are met, and have assets of roughly $141,852,000 (ignoring any cash flows).
If the $100-million fund gets a repeating 26% and -14% return pattern, after six years it has a deficit of around $14,617,000 and is roughly 90% funded.
We clearly need to take account not only of what return might be achievable in any given year but also how much volatility in the return there is likely to be from one year to the next.
Pension funds are a pay-me-now-or-pay-me-later proposition. If our return expectations are always lagging reality, resulting in decades of downward adjustments in reaction to negative investment experience, the net effect is a transfer in wealth from one generation to the next. If insufficient funds have been set aside to fund the pension promises given to older workers, then younger workers will have to make up the difference, either directly through increased contributions or decreased pensions, or indirectly through lower wages as funds are diverted to pay legacy costs.
For a pension system to be fair, the assumptions would have to be unbiased.