Release of Australian Life Table 2020–22

As this current investigation spans three years where mortality experience was impacted by the COVID-19 pandemic, I thought it worth setting out our approach to the establishment of the ALT, some of the findings and a comparison to the life table published by the Australian Bureau of Statistics (ABS). This article is deliberately brief and I refer the reader to the publication for a fuller discussion of ALT2020–22.

Approach to dealing with the COVID-19 pandemic

The ALT’s primary purpose is to provide an historical record of the mortality experience of Australia during the period of investigation. The ALT has not been adjusted for the effects of the COVID-19 pandemic (or any other cause of variation in this period).

However, for this publication, we added a new section showing how the crude mortality rates for each individual year compare with the average over the three-year investigation period. We provided this information because some users may wish to further understand the impact of year-to-year variations when developing their assumptions in respect of future mortality.

Consistent with the experience of excess mortality reported during the COVID-19 pandemic, this new section shows that the underlying mortality experience was quite different in each year.

2020 and 2021 generally experienced mortality that was lower than the three-year average; only a small number of (mostly younger) age groups had higher-than-average mortality and the 2020 experience for ages 65 and older was particularly low. Conversely, 2022 experienced mortality that was materially above the three-year average at most ages. There are two drivers of the variance observed from year to year:

1. The absence of influenza in Australia in 2020 and 2021 due to the measures introduced to curb the spread of COVID-19. This resulted in ‘negative’ mortality displacement, with some deaths that would normally have occurred in 2020 (if not for the pandemic) likely occurring in 2021 and 2022 instead.
2. Relatively widespread COVID-19 in 2022 once pandemic restrictions were lifted and borders re-opened.

We note that year-to-year variations are not confined to the current investigation. For example, the same information for ALT2015–17 is expected to have shown the 2017 year materially higher than the three-year average due to the impact of the adverse influenza season on mortality in that year.

Changes in mortality rates since 2015–17

Figure 1 shows the mortality rates from ALT2020–22 together with those reported five years earlier (ALT2015–17). It shows that mortality rates have fallen for most ages. There are three exceptions where mortality rates increased slightly:

1. very young children
2. those nearing and in their 20s
3. those at very advanced ages.

The increase in mortality rates at advanced ages has been observed, albeit to varying degrees, in each ALT since 2005–07.

Figure 1: Mortality rates, 2015–17 and 2020–22

Period life expectancies

The period life expectancies resulting from the current and previous ALTs are shown in Table 1. Despite the COVID-19 pandemic, both sexes have experienced an improvement in life expectancy over the five-year period of around half a year.

In other words, the substantially higher mortality experience of 2022 was not so great that it eliminated mortality improvement entirely.

Table 1: Period life expectancy at birth (years) – ALT2020–22 compared with ALT2015–17

Table 2 shows a comparison of life expectancy at birth from the current ALT with the projected life expectancy from ALT2015–17 if the 25-year and 125-year mortality improvement factors had applied for the period 2016 to 2021 (but without future mortality improvement).

The current life expectancy is lower than both projected life expectancy measures, and close to the projection using the 125-year mortality improvement factors. This suggests that the overall impact of the COVID-19 pandemic on the three years covered by the investigation was to slow the rate of mortality improvement that may otherwise have occurred, again consistent with reported excess mortality across those years.

Table 2: Period life expectancy at birth (years): ALT2020–22 compared with projected from ALT2015–17

I note that period life expectancy as a single summary statistic cannot provide information on the diversity of outcomes.

For example, under the mortality rates reported in the current ALT, around 60 per cent of the population would be expected to survive beyond the reported life expectancy. This result is separate from the issue of mortality improvements that might occur over an individual’s life as discussed in the following section.

Mortality improvement factors

The publication includes historical rates of mortality improvement and discussion of how mortality has improved over time for selected ages. The trends differ quite markedly depending on the age in question. I have provided two measures – mortality improvement over the past 125 years and over the past 25 years. Cohort life expectancies are also provided for illustrative purposes, where I have adopted the historical 125-year improvement factors for all future years.

I remind practitioners that the adoption of future mortality improvement factors is a matter of actuarial judgment. The circumstances of the situation in which they are being used needs to be considered when forming these judgments (as for any actuarial assumptions). For those who may want to perform their own analysis of mortality improvement, I have provided a spreadsheet containing mortality rates from all previous ALTs.

Comparison with the ABS Life Tables

The ABS also produces life tables that are published annually (whereas the AGA ALTs are only published every five years). I am aware that practitioners often adopt the ABS life tables rather than the most recent ALT, hence the inclusion of the comparison here.

Broadly, the approach used by the ABS is similar to the AGA approach, however there are some differences:

• The ABS releases their life tables in early November of the year following the end of the three-year investigation period (i.e., the 2020-22 life tables were published in November 2023). Given the relatively short period between the end of the investigation period and the publication date, the ABS is required to estimate the number of late reported deaths. ALT2020–22 uses data that is extracted around one year later than the ABS, thus an estimate of late reported deaths is not required.
• The graduation methods used differ. The ABS smooths the crude death rates through the application of the Hodrick-Prescott filter for most ages, with additional methods used for very young (0-1 years) and very old (95-120 years) ages. The three methods are then blended to ensure smooth transitions between the models.

Figures 2 and 3 show comparisons of the ABS mortality rates and ALT2020–22.

Figure 2: Mortality rates, ABS and ALT2020–22

Figure 2 shows that the mortality rates produced by the ABS and the ALT2020–22 rates are similar and, in most cases, not visibly different.

Figure 3: Percentage difference in mortality rates (ALT2020–22/ABS)

Figure 3 shows that the percentage difference in the rates below age 20 vary somewhat (mostly by around +/-7%) but with neither series consistently higher/lower than the other.

For ages 20 to 45, the rates are closer (differing by around +/-1.5%), and again with neither series consistently higher/lower than the other. However, between ages 45 and 80, the ALT2020–22 mortality rates for both males and females are consistently lower than the ABS rates. The rates are then similar between ages 80 and 90. After age 90, the ALT2020–22 rates are again lower than the ABS rates, and substantially so.

These differences in mortality rates produce differences in life expectancies at birth, at age 65 and at age 90 as shown in Table 3.

Table 3: Period life expectancy (years) – ABS Life Tables compared with ALT2020–22

The differences in the graduated mortality rates in the two life tables result in a small difference in life expectancy at birth for males and no difference for females.

At age 65, ALT2020–22 results in life expectancies for both males and females that are 0.1 years higher than the life expectancy resulting from the ABS life tables.

At age 90, ALT2020–22 results higher life expectancies than the ABS life tables for males of 0.2 years and 0.1 years for females.

While the number of years difference here is similar to that at age 65, the difference is proportionately much greater (as those aged 90 have a future life expectancy of 4.5 to 5.0 years, whereas those aged 65 have a future life expectancy of 25 to 28 years).

CPD: Actuaries Institute Members can claim two CPD points for every hour of reading articles on Actuaries Digital.