Until this year, no new mortality tables had been published by the Canadian Institute of Actuaries (CIA) since 2010. This change, the development of table CIA2014, useful for those working in pricing, valuation and experience research, reflects the fact that mortality experience has changed over time, says researcher and actuarial consultant to the CIA, Bob Howard. 

“The most recent CIA individual life mortality study concluded that there is a need for a new table because neither CIA8692 nor CIA9704 have a slope by age or duration that fits well with recent experience,” the report accompanying the table states. 

CIA2014: A Mortality Table Constructed from the CIA Individual Data of Policy Years 2009-2019, is based on 10 years of data taken from an annual mortality study involving multiple insurance companies which has been ongoing for 71 years. The annual intercompany mortality study, the CIA says, has been important to establish a benchmark for industry experience. For those who wish to study the data further on their own, a .csv database is available to download from the report. 

In constructing the table, researchers found no mortality improvement for female smokers. This was surprising as mortality improved for every other demographic and subset. Policies issued due to conversion, guaranteed insurability, or with a substandard rating are excluded, as are guaranteed issue and joint policies. 

Does not reflect COVID-19 deaths 

The table, constructed using data from 2009 to 2019, also does not reflect COVID-19 deaths. “The CIA will need to monitor mortality experience more closely than usual over the next few years to determine whether there is any long-term impact on mortality from COVID-19,” the reports states. 

Given that the mortality tables reflect insurance experience too, it is also not directly reflective of trends in the general population. The opioid epidemic, for example, is not extensively reflected. Howard estimates that probably between 20 and 30 per cent of the population is insured. “To the extent that people have bought insurance and have later fallen prey to the opioid (epidemic), they are in the data. But I can tell you, we don’t see a lot of those claims.” 

Similarly, he agrees that insured individuals do appear to live longer, but says the link is not causal – the selection which occurs in insurance purchasing removes from the sample those who would not be insurable. “If you look at the averages of all those who are insured, by and large they do live longer than those who are not insured. One reason is that the general population would include people who have, for example, very serious congenital problems and people who were born and who died on the first day of life. They will never get insured, but they will be counted in the population,” he says. 

Scant experience data at very high ages 

What the table does show is the somewhat scant experience data that companies have to work with at very high ages. 

“Rates at the highest ages are important to the table, but experience at those ages is not only scant, but unreliable,” the report’s authors write, adding that rates at the oldest ages are extrapolated. 

“In the past, members have expressed a need for the selection portion of published tables to extend to issue age 90. The experience over age 80 is sparse. Consequently, the rates over age 80 must be an extrapolation and they must be used with caution.” 

The issue arises because the industry has no requirement for reporting all deaths the same way the federal government tracks death counts. A client purchasing a policy in the 1950s with a small policy that has been paid up for years, could die without the insurance company ever finding out. Howard says it’s not a high proportion, but a small percentage of deaths do generally go unreported when clients are over 90. “By the time we get to age 100 it starts to have a material impact on mortality rates,” he says. “It’s a problem we’ve known about for some time. We adjust for it.” He says this works fine, but it is something the data’s users do need to be aware of. 

Although this data fact is unlikely to have much impact on pricing, he adds that it does impact those who work on evaluating reserves companies need to keep in hand to cover their existing liabilities. “My example – a $5,000 policy bought in the 1950s – is a very small policy, but if there are enough of those, it’ll make a difference,” Howard says. “If you don’t have an updated table, you’re not necessarily going to be holding the right amount.” 

Similarly, “smoking unknown” mortality rates – these reflect those who purchased policies before the early 1980s and those today who are very young and not asked about their smoking status – can be used to pretty accurately calculate required reserves, “but nobody other than the very young are smoking unknown for new issues. It’s really, largely irrelevant for people doing pricing, but it is important for those who are doing evaluations.” 

Finally, he says the updated tables were not constructed to replace the older ones because they are faulty. Instead, he and the report both point out that mortality simply changes over time.

Overdue for an update 

“You could have a 100 per cent accurate mortality table but it it’s 20 years old, it’s almost useless. That’s the situation. We were overdue for an update,” he says. “I think it reflects very well the experience that is underlaying the table.” 

The sentiment is echoed in the table’s accompanying report: “Mortality improvement is a fact of life in Canada,” he writes. “Although there are fluctuations at various times and ages, the general trend of mortality is downward and had been for many years. It seems reasonable and desirable to reflect mortality improvement in the construction of a (new) mortality table.” 

While the impact on pricing overall will not be huge – Howard points out that larger companies don’t wait for tables to be updated, they have their own statistics to draw from – he adds that the impact won’t be non-existent, either. 

“Mortality rates improve generally, but they don’t improve at the same rate for all ages, for both sexes and for smokers and non-smokers. It’s always changing. This will bring things up to date,” he says. “It will be particularly important for the smaller companies that can’t rely on their own experience.”