Everyone knows why organizations around the world have been closing down their DB pension plans in recent years: in a Sophie’s Choice of sorts, they have opted to reduce risk on behalf of shareholders rather than absorb risk on behalf of employees. A 2003 Morneau Shepell survey confirmed this. The question is whether or not there is a more tangible way to express that risk.
Consider a DB plan with an employer cost of 9% of pay on an ongoing basis (i.e., the normal cost). Employers might think that a DC plan benefit of equal value would also cost them 9% of pay, but, in fact, it would be more. To figure out how much more, we can look to the capital markets for guidance.
Imagine being promised a stream of payments of, say, $100 a month over the next 20 years. The present value of those payments depends on the creditworthiness of the payer. If the payer is the federal government, the value would be about $17,300 (assuming an average spread between long Canada bonds and single-A bonds of 1.3%). If the payer is a single-A corporation, the value would be about $15,500, or 11% lower. These numbers are based on the long-term average spread between Government of Canada bonds and single-A corporates. In recent years, the gap is wider still, but I will use the 11% figure here for the sake of conservatism.
Over long periods, the cumulative default rate on single-A corporate bonds with a term to maturity of 20 years or more has been a shade under 5%. If default meant forfeiting all 20 years of payments, then the expected loss on a single-A promise would be about 5%, being the default rate multiplied by the forgone payments. Since the default rate is a cumulative figure over the entire 20-year period, however, only half the payments—on average—would be affected by a default, and even then some value can usually be recovered. Consequently, the expected loss on a single-A promise is closer to 2%.
Hence, the market discounts a single-A corporate promise by 11%, even though the expected loss is only 2%. The introduction of even a minimal amount of uncertainty (i.e., a one in 20 chance of default with a 2% expected loss) is enough to reduce the value of a single-A promise by another 9%.
We now have an idea of just how much the markets punish uncertainty. Let’s apply this same thinking to pensions. The volatility in the retirement income stream under a DC plan is going to be at least as high as the volatility of income from a portfolio of single-A bonds. In fact, the volatility will probably be much higher since employees have to invest a sizable portion of DC assets in equities to gain a decent return. Rounding up, a DB plan with a 9% normal cost is equivalent to a DC plan with an employer contribution of 10% or more. This is where we find a disconnect between theory and practice.
Morneau Shepell’s database of hundreds of DB pension plans shows that the average employer normal cost within these DB plans is about 8.5% of pay. Using the above analysis, the average employer contribution to DC plans should be roughly 10% of pay. In fact, the average employer contribution to mid-size DC plans in Canada is only 5% of pay. (Explaining this gap is the subject of another article.)
I note that the gap is even wider than suggested above since I am assuming no “leakage” due to the various inefficiencies inherent in a DC plan. DC plans are a less efficient retirement vehicle than DB plans for a variety of reasons. First, they pay out more money to employees who leave mid-career, leaving less to those who stay until retirement. Second, DC plans will overshoot the retirement income target for some employees from time to time—when markets are more sanguine than we have seen lately. Finally, DC participants tend to make poorer investment decisions, which is not surprising considering that a 2001 Morneau Shepell employee survey shows that fewer than half of participants even know what a money market fund is.
The conclusion seems apparent. Either DC plan sponsors should generally be contributing more or DB sponsors should be offering less. This is especially true within industry sectors where some competitors provide DC and others provide DB.