Mortality meets volatility in pension market

“In this world nothing can be said to be certain, except death and taxes.” America’s first blogger Benjamin Franklin made this observation two and a quarter centuries ago and it has held true these many years.

With the latest developments regarding mortality assumptions, however, Franklin may want to consider a rewrite… (if he wasn’t, you know, certainly dead.) For the new world of pension mortality has made death anything but certain.

Until recently, most mortality tables were “static,” meaning there were no explicit projections of future mortality improvements. These static tables were updated about once a decade, so mortality assumptions in most years introduced little uncertainty to liability calculations.

The Pension Protection Act of 2006 greatly accelerated the use of simple “generational” mortality assumptions. Generational assumptions are actually a combination of a base table and an improvement scale. They, too, are very predictable from year to year between updates of the underlying data.

A base mortality table assigns a probability of death at each age. While an improvement scale determines how quickly future years’ death probabilities will decline (usually), effectively estimating the impact of future longevity improvements. The longer a pensioner is assumed to live, the more benefits they will be expected to receive, and the higher the associated liabilities recognized by the plan sponsor.

The source of recent death uncertainty is a paradigm shift in longevity thought contained in the Society of Actuaries’ Retirement Plans Experience Committee (RPEC) release of the “MP-2014” improvement scale in October 2014.

Culminating five years of analysis and research, MP-2014 delivers an impressive array of historical mortality improvement data distilled into two-dimensional, multi-colored “heat maps.” These quantitatively hallucinogenic exhibits are extremely difficult to interpret, as shown here:

Generally, however, they indicate mortality improvements leading up to 2007 (the last year of available graduated data) were quite strong. The historical heat map results are then extrapolated gradually over a 20-year “convergence period” toward a long-term improvement assumption of 1% for most ages relevant to pension valuations.

This 20-year convergence period places a lot of emphasis on recent mortality experience. Theoretically, a mortality improvement blip remains in play for two decades, adding real liabilities for what may just be a statistical anomaly.

“Mort-ility” is born    

In addition to heat maps, RPEC also included an intention to update data for emerging historical mortality experience at least every three years, perhaps even annually. So if you’re keeping score at home, MP-2014 moves away from simpler, less sophisticated improvement scales that change infrequently in favor of:

The combination of these two factors greatly increases the uncertainty of mortality’s impact on plan liabilities from year to year. Now sponsors already contending with investment and interest rate volatility are faced with another challenge, one that can be neither avoided nor hedged: mortality volatility.

Recognizing that there are very few phrases in the English language harder to utter than “mortality volatility” (say it 10 times fast), I will hereafter refer to this phenomenon using the newly coined term, “mort-ility”!

A number of actuaries expressed mort-ility concerns (though they didn’t call them that) during the MP-2014 comment period. They suggested shorter convergence periods may be more appropriate and would introduce much less mort-ility, as the impact of new data would work itself out of the long-term improvement assumption more quickly.

Reasonable alternatives

Much to their credit, RPEC accepted the comments of the actuarial community that the original MP-2014 exposure draft was “overly restrictive.” Allowances for reasonable alternative assumptions using the RPEC model were included in the final draft, giving actuaries and plan sponsors some much-needed flexibility.

Some actuarial providers (including my company) have designed their own standard mortality improvement scales that satisfy RPEC allowances, most of which use convergence periods much shorter than 20 years. If accepted as reasonable by plan auditors, these alternatives have the potential to significantly reduce mort-ility.

Source: Employee Benefit Adviser