Life expectancy data that was expected to add €10bn to the liabilities of German Direktzusage pension arrangements was flawed, according to the actuarial consultancy that produced the numbers.

Longevity improvements were overestimated in revised data tables issued by Cologne-based actuarial consultancy Heubeck over the summer.

The company – which has been the official provider of longevity figures for the German pension industry for decades – said it discovered “inconsistent data” was used for calculating the latest tables.

In July it issued a new set of longevity tables (known as RT 2018 G) showing a continued – albeit slower – increase in the average life expectancy in the country.

The new calculations would have triggered companies offering Direktzusage pension plans to increase their reserves by 1% under national accounting standards, equivalent to roughly €10bn. Direktzusage plans are benefits paid directly from companies’ balance sheets, rather than from separate funds.

Heubeck conceded that it “slightly overestimated the trend to further increases in life expectancy”.

“This means pension reserves calculated on the basis of the RT 2018 G would be higher than those based on a consistent deduction of the mortality trend,” the consultancy said in a press release.

For the first time in the 2018 revised tables, Heubeck had applied “socio-economic factors” to its calculations to adjust for the modern lifestyles of pensioners.

In its statement, the consultancy noted that, while these factors had contributed to a higher increase in longevity and “might still make sense under international accounting standards”, the tables would be revised.

“We are sorry the inconsistencies have not been noticed before,” Heubeck added.

Speaking to IPE in August, Georg Thurnes, chief actuary at Aon Hewitt Germany, raised concerns about the application of socio-economic standards as he would rather have applied them to the active phase of employment. 

The revision of the revised tables is expected to be finalised by mid-October.