Presentation at the SOA Living to 100 International Symposium on the topic of: Mortality Projections for Social Security Programs in Canada. 2017 Living to 100 Symposium, General Session IV, Orlando, FL, 5 January 2017

Good afternoon. I am Annie St-Jacques, Actuary at the Office of the Chief Actuary in Canada. I am pleased to be here to talk to you today with Jean-Claude Ménard, Chief Actuary of the Canada Pension Plan, the Old Age Security Program and federal public sector pension plans in Canada.

Presentation Outline (Slide 2)

Here is a brief outline of the presentation we have for you today. I will first talk about how we develop the population mortality projections used for the most recent Canada Pension Plan Actuarial Report. In developing our assumptions we examine the past, estimate which past trends could continue into the future, and look at emerging trends. I will then pass it over to Jean-Claude who will compare Canadian and US mortality rates. He will also show you the probability of living beyond age 80, 90 and 100 and he will even provide some examples of what is mathematically needed to achieve a life expectancy at birth of 100. He will conclude with the long-term financial sustainability of the CPP.

Life Expectancy at Birth and at Age 65 (Slide3)

This slide presents the calendar year life expectancy at birth and at age 65. Life expectancies at birth are the two lines in the upper part of the chart and are associated with the left axis. It has increased steadily since the beginning of the 20th century. However, we can observe that the slope becomes flatter. In the first half of the 20th century, life expectancy at birth increases by 19 years for males; it represents an average pace of increase of around 4 years per decade. In the second half of the 20th century, life expectancy at birth for males increases by 10 years, representing an average pace of increase of 2 years per decade.

Inversely, for life expectancies at age 65, the two lines in the lower part of the chart associated with the right axis, the significant growth is relatively recent, especially for males. From 1945 to 1975, life expectancy at age 65 for males was very stable, between 13 and 14, while it increases by 5 years since 1975. For females, a continuing increase has started in the 50’s.

The gap between female and male life expectancies has increased from 1920 to 1980, but has started to narrow thereafter.

Contribution to increase in life expectancy at birth has gradually shifted to people over age 65 – Males (Slide 4)

We just saw that there is a slowdown in the rate of increase of life expectancies at birth between the first and later parts of the 20th century, while the growth in the rate of increase of life expectancies at age 65 is more recently observed. This table provides more details on the change in life expectancy over the past 80 years broken down by 20-year periods. It shows the total increase in life expectancy for males for each 20-year periods and from which age category the change comes from. It can be observed that the proportion of the increase in life expectancy coming from the reduction of mortality rates at ages 65 and over becomes more and more important over time. Over the last 20-year period, from 1991 to 2011, 58% of the increase in life expectancy for males came from the reduction of mortality rates at ages 65 and over. This trend is expected to continue in the future.

Contribution to increase in life expectancy at birth has gradually shifted to people over age 65 – Females (Slide 5)

The same trend is observed for females. The percentage of the increase in life expectancy coming from reduction of mortality rates at ages 65 and over is even higher, standing at 65% over the 20 years from 1991 to 2011.

Life expectancy is impacted by level of income and marital status (Slide 6)

We know for sure that lifestyle affects life expectancy, maybe even more than genetics, but we don’t have the information to quantify their impact. Same thing for enhanced medical technologies. However, from the experience data we have on Old Age Security beneficiaries, what we can quantify is the impact of some socioeconomic factors as level of income and marital status.

The Old Age Security Program in Canada covers all Canadian population. The Old Age Security mortality study indicates that married beneficiaries tend to live longer than the overall population, while single beneficiaries have a shorter life expectancy, with males showing a significant difference of 2.4 years. The same study indicates that beneficiaries with higher levels of income (meaning those not entitled to the income-tested Guaranteed Income Supplement benefit, the GIS) have a higher life expectancy than the overall population. This observation may be explained by the relationship between higher levels of income and improved health and quality of life.

On the left side you have the information on the impact of marital status. On the right side, you have the information on the impact of level of income. Now, if we combine this information, what do you think is the overall impact. Who is living longer? A married male with GIS or a single male with higher income (not receiving GIS). It appears that marital status plays a more important role than income for men. Married males with GIS lives almost one year longer than single males not receiving GIS. At age 65, the remaining life expectancy of a single male not receiving GIS is 17.7 years while it is 18.6 years for a married male with GIS. For women, the marital status is of similar importance than the income. In other words, it almost balances each other. At age 65, the remaining life expectancy of a single female not receiving GIS is 21.8 years while a married female receiving GIS has a remaining life expectancy of 21.4 years.

Improvements in mortality related to heart diseases have been significant over the last 15 years (Slide 7)

Looking at mortality rates by cause of death for males and females over age 65, we can see that the significant increases in Canadian life expectancies observed over the last few decades is in great part due to the improvements in mortality related to heart diseases. These rates were improving at around 4% per year.

Cancer is now the cause of death with the highest mortality rate. Reductions in mortality from cancer may hopefully become an important factor in the future.

CPP-OAS Average Annual Mortality Improvement Rates [males] (Slide 8)

At the time we determined the assumption for the last Actuarial Report on the Canada Pension Plan, the information from the human mortality database was available up to 2011. However, with the experience data we have on Old Age Security and Canada Pension Plan beneficiaries, we were able to analyze more recent trends. As shown in those charts, OAS and CPP average annual mortality improvement rates for males in age groups 65 to 74 and 75 to 89 have been lower over the most recent 5-year period from 2010-2015. The OAS and CPP average annual mortality improvement rates for males in the age group 65 to 74 (top part of the slide) have gradually declined steadily over the last three 5-year periods. For the age group 75 to 79 (bottom part of the slide), mortality improvement rates have increased between the period 2000-2005 to the period 2005-2010, but then decreased in the next period. It is not shown here but for each of these two age groups, female mortality annual improvement rates have generally been more stable and lower compared with males. However, they have experienced a decrease in mortality improvement rates from 2005-2010 to 2010-2015.

The ultimate mortality improvement rates are sometimes based on historical averages (Slide 9)

So, while we may make an educated guess on how the mortality rates will evolve in a short term, how to make assumption on the ultimate mortality improvement rates? Looking purely backward, we can see that over the last 90 years (let’s take the longest available period since our projections are very long-term), the average MIR for Canada were 0.9% per year for ages 65 and over and 0.6% per year for ages 85 and over.

Estimates of MIRs for ages 65+ for 2012-2014 incorporate OAS experience (Slide 10)

Here we have the average mortality improvement rates for age group 65-74 for males in blue and females in red. It shows a portion of the historical data and the projection. For the short term, 2012-2014, as mentioned earlier, we considered the decreasing trends observed in mortality improvement rates for OAS beneficiaries. The ultimate rate reached in 2032 is 0.8% for both males and females. It is based on the analysis of the past Canadian trends but there is also judgement on what extent mortality can be improved in the future considering analysis of possible drivers of future mortality improvements. For the convergence period, between 2015 and 2031, it follows a cubic function based on the CMI Mortality Projection Model. [CMI stands for Continuous Mortality Investigation]

Males Mortality Improvement Rates 15-year Average (Slide 11)

Heat map is a useful tool to analyze the trends in mortality improvement rates. This analysis is usually performed on smoothed mortality improvement rates. Mortality improvement rates for any given age, sex, and calendar year may be regarded as a combination of age, year, and cohort components or effects. Age effects are seen as horizontal bands or patterns on heat maps, calendar year effects as vertical patterns, and cohort effects as diagonal ones.

This one is the historical heat map for Canadian males. It is the historical 15-year moving average based on CHMD data, for all ages and several past decades. The red represents high improvements (between 4% and 4.5%); the orange represent smaller improvements, in the next 0.5% range; then comes the yellow, followed by two different shades of green, then blue, etc.

As it could be seen, in late 60s – early 70s, the mortality rates among young males increased significantly (purple spot). We believe that this phenomenon was caused by increase in accidental deaths caused in particular by unsafe driving. Deterioration in male mortality could be observed in mid-90s for males aged between late 20s and early 40s and is related to AIDS. Another interesting phenomenon that could be seen on this slide is the cohort effect for males born approximately in the 1930s and 1940s. These cohorts experienced higher mortality improvement rates than those born before or after.

The part on the right of the dotted line represents our projection of the MIR. The light orange area between age 65 and 75 is now closing and is being replaced by yellow.

Females Mortality Improvement Rates 15-year Average (Slide 12)

For Canadian females a significant calendar year effect is seen for aged less than 45 in the 1950s and early 1960s with mortality improvement rates greater than 5% per year. It is believed to be related to reduction of mortality as a result of giving birth. The recent historical improvement rates are more moderate.

Jean-Claude will now take over the presentation.

For ages 65 to 74, 7 deaths per 1,000 are from cancer, while only 3 deaths per 1,000 are from heart diseases (Slide 13)

For the next few slides, I will compare evolution of mortality rates by age groups for Canadians and Americans with more focus on Canada. United States numbers are based on the 2015 OASDI Trustees Report.

For age group 65 to 74, the recent reduction of 52% in mortality rates over the last 40 years was much more significant compared to 22% reduction over the previous 40-year period. The reason is that while female rates decreased steadily, male mortality rates for this age group have been decreasing at a much faster pace in the last two decades than in previous decades. A further reduction of 40% is projected. For this age group, cancer is responsible for the vast majority of deaths. Current mortality is 22% lower than for U.S. mainly due to much lower mortality caused by diseases of the heart and lower respiratory diseases.

Male mortality rates for ages 75 to 84 for Canada are projected to become similar to US female mortality rates (Slide 14)

Age group 75 to 84 has experienced similar pattern and magnitude of reduction in mortality as age group 65 to 74. Current mortality is 17% lower than for US, again, mainly due to lower mortality caused by diseases of the heart rates and lower respiratory diseases. Significant reduction in age-specific mortality rates has contributed to the aging of the population. For the age group 75-84, you can observe that mortality rates have continually decreased over the last 80 years. The reduction was about 45% over the last 40 years compared to only 22% over the previous 40-year period. A further reduction of 39% is projected. Mortality rates for canadian males are projected to become the same as U.S. females.

Elderly mortality has decreased over the last 80 years, more so over the last 10 years (Slide 15)

U.S. and Canadian mortality rates of the 85 to 89 age group were quite similar prior to 1999. However, since 1999, Canadian rates have been reducing faster than American ones. The reduction over that decade was 22%. Canadian rates are currently 10% lower than for U.S. mainly due to much lower mortality caused by diseases of the heart and Alzheimer’s. A further reduction of 35% is projected by 2051.

Three-quarters of Canadian men aged 20 today are expected to live to age 80 [82% of women] (Slide 16)

So, what are our chances to live to advanced ages? This slide presents the probability of surviving to age 80 from a given age in 2015 for Canada (in red), Switzerland (in black), the U.K. (in blue) and the U.S. (in green). It is interesting to look at the shape of the curves. It will be even more pronounced on the next slide showing the probability of living to 90. The probability is slightly higher for younger ages due to the projected decreases in mortality rates. On the other hand, for older age groups, the probabilities of living to 80 increases since only individuals who have already reached older ages are considered. As it could be seen, for all four countries and for all ages, chances to survive to age 80 are quite high. For example, according to the CPP27 projections, 75% of Canadian men and 82% of Canadian women aged 20 today are expected to live to at least age 80.

Near half of Canadian men aged 20 today are expected to live to age 90 [58% of women] (Slide 17)

Living to 90 on average is also quite a realistic perspective. Again, this slide presents the probability of surviving to age 90 from a given age in 2015 for Canada, the U.K., Switzerland and the U.S.A.

According to the CPP27 projections, almost half of Canadian men and 60% of Canadian women aged 20 today are expected to live to at least age 90.

8% of Canadian men aged 20 today are expected to live to age 100 [14% of women] (Slide 18)

Now… can we live beyond 100? According to the CPP27 projections, 8% of Canadian men and 14% of Canadian women aged 20 today are expected to live to age 100. UK has the highest probabilities of living to 100 of the four countries at all ages between 0 and around 85 due to a higher assumed future mortality improvement rates.

Uncertainty of Results: Life Expectancies at age 65 if MIRs by cause are sustained (Slide 19)

Does someone believe that men will outlive women in the future? Future mortality projections are very uncertain. As such, it is useful to consider alternative projection scenarios. This chart presents the life expectancies at age 65 under a scenario where the annual MIRs of the last 15 years (1998-2012) by cause are assumed to remain constant over the entire projection period.

Due to the higher recent mortality improvement rates for males, especially for death related to cancer, this scenario leads to a narrowing of the gap between male and female life expectancies at age 65 and, eventually, higher life expectancy for males than for females by 2043 and thereafter. In 2075, male life expectancy would surpass the one of females by 1.6 years. This situation is quite unprecedented for mammals. This scenario serves as a reminder that setting future assumptions only on recent experience may lead to unintended results.

I want to stress that even if the increase in life expectancy is quite important under this scenario, people would not live beyond 100 on average. Males would have a life expectancy at age 65 of 30 years, which led them to 95 years, still 5 years from 100.

Uncertainty of Results: Life Expectancies at age 65 if mortality from cancer is wiped out (Slide 20)

Another scenario could be to gradually remove mortality from cancer over the next 75 years. This is quite unrealistic but demonstrates that even such an ideal world would not increase the life expectancy beyond 100 in 2075. Both males and females would gain approximately 2 years.

Survival Curves for a Life Expectancy of 100 [Males] (Slide 21)

What should happen to mortality rates in order for Canadians to have a life expectancy at birth of 100? We came up with few alternatives. This chart presents a comparison of the current survival curve and two alternative ones. Under the first scenario, current mortality rates at each age are reduced by 84% (dotted blue survival curve). Under the second alternative, the life-span is increased to 140 (red survival curves), and the current mortality rates are redistributed across ages 0 to 140. The third alternative is very simple – nobody is assumed to dies before age 97, and then current mortality would apply.

To live beyond 100… (Slide 22)

If we wish to live to 100 today, we either need to reduce current mortality rates dramatically or increase our life span. In our view, both routes are not very likely. In respect of the increasing life span, a paper published in Nature Journal of Science last October argues that the maximum lifespan of humans is fixed and subject to natural constraints. The maximum reported age at death is generally set at 122 years, the age at death of Jeanne Calment, still the oldest documented human individual who ever lived.

According to our projected mortality improvement rates, if mortality rates decrease at the same pace as observed over the past 15 years, a life expectancy of 100 would be attained only after 2200. If mortality rates decrease at twice the pace observed over the past 15 years, a life expectancy of 100 would be attained in about a century.

So, what is the impact of living longer on the CPP? (Slide 23)

So, why worry about life expectancy when it comes to the CPP? The reason is the uncertainty embedded in mortality projections.

Life expectancy at age 65 in 2016 is 21.3 years for males and 23.7 years for females, including the future mortality improvement. However, the same measure in 2050 is 23.3 years for males and 25.6 years for females. This means that, based on our assumptions of future improvement in mortality rates, a new CPP beneficiary aged 65 in 2050 can expect to receive his/her pension for about 2 years longer than a beneficiary in 2016. The challenge is that there is high uncertainty related to future mortality improvements.

A low and a high cost scenarios are presented here. In the low cost scenario, mortality is assumed to improve at a slower pace than expected. Mortality improvement rates gradually decrease to an ultimate value of 0% in 2032 instead of 0.8%. This scenario would result in life expectancies 2.4 years lower than the best-estimate. The CPP minimum contribution rate would go down by 0.3 percentage point, from 9.79% to 9.46%. In the high cost scenario, mortality is assumed to improve at a faster pace than expected. The ultimate value of the mortality improvement rate is doubled compared to the best estimate and corresponds to 1.6%. This scenario would result in life expectancies around 2.5 years higher than the best-estimate. The CPP minimum contribution rate would increase to 10.10%.

If no future mortality improvements are assumed at all after 2011, projected life expectancies would remain at their 2011 calendar year value for all future years and the minimum contribution rate would decrease to 8.99%. The difference of 0.8% with the minimum contribution rate represents the cost of improving longevity.

Conclusion (Slide 24)

To conclude:

  • Over the next 20 years of projection, 69% of increase in life expectancy at birth for men will come from the reduction of mortality rates past age 65. For women, this proportion is 76%.
  • Projection of mortality rates is a difficult exercise and future mortality rates are highly uncertain, especially for people older than age 90.
  • Despite increased longevity of Canadian population, the CPP is expected to be sustainable over the long term based on the most recent Actuarial Report tabled before Parliament on Sept. 27, 2016.

Thank you. I will be pleased to answer any questions you might have.