Pension liabilities heavily depend on forecasts of the mortality of the pension plan beneficiaries. These forecasts are produced by actuaries. Until quite recently, the actuarial science was deemed to be capable of coming up with a reliable point estimate (if you can speak of point estimates in the context of large cohort mortality tables) of future mortalities. This expectation does not survive under closer scrutiny of course. That is not a bad thing, because point forecasts of the future are necessarily inaccurate, therefore such expectations were never realistic.
The British actuarial profession is leading the pack again with its publication of a draft library of mortality projections, indicating that it may be reasonable to utilise a number of scenarios in mortality projections. But more closely to home, and indicative of imminent changes to current mortality assumptions and thus liability valuations, they warn that currently employed mortality tables may considerably underestimate improvements in future mortality. We understand this to imply that more realistic tables will assume higher average life expectancies and, consequently, higher pension liabilities.