Beyond life expectancy: A different approach to comparing the impact of life saving measures on future demographics

Making accurate predictions about population trends is difficult. However a good grasp of demographic trends is also essential both for policymakers and many companies. Changes in Iife expectancy is a relatively blunt tool for measuring the impact of interventions, such as a new drug, or safety law, on different causes of death. Séverine Arnold and her co-authors offer a more sophisticated approach which enables interventions to be measured in terms of their relative impact on the age distribution of future populations, including old age dependency ratios.

Having a good understanding of demographic trends is very important for many organizations. For life insurance and pension firms where pricing, risk management and ultimately profitability, are closely linked to demographics. For all those companies that target specific segments of the population with particular products and services. And for governments too, where poor demographic forecasting can lead to policymakers designing policies that produce unforeseen, unintended, unwanted consequences.

The challenge lies in the complexity of demographics. Making accurate predictions about populations is problematic. Fortunately, though, academics such as Séverine Arnold, and her co-authors Alexandre Boumezoued, Nicole El Karoui and Héloïse Labit Hardy, are making considerable progress in advancing our understanding. To do so the team make innovative use of models developed primarily for studying tree and bird populations, instead applying them to human populations.

Previous studies have investigated the impact of causes of mortality on life expectancy. However, in her co-authored paper, “Cause-of-Death Mortality: What Can Be Learned From Population Dynamics?” Arnold takes a step further, combining mortality analysis with population dynamics to provide a more sophisticated analysis of how causes of death affect the age distribution in the future.

For example, how might curing a particular disease, or reducing cycling deaths by making cycle helmets mandatory, affect the age structure of the population over time? And, more specifically, as a result of the impact on age distribution how might that affect the old age dependency ratio. This is a ratio which takes the number of people 65 years and older and divides that older population by the number of active population aged between 15 and 65 years old. It is a ratio that provides an idea of the burden of the old age population on the active population; important for social security system planning, for example.

A particularly interesting aspect of Arnold’s approach was a focus on comparing the effect of restricting deaths from different causes (overall the team looked at six different causes of death). Life expectancy is the metric we tend to think of when gauging the impact of a specific cause of death. What impact does a policy have on life expectancy, for example? So if you were considering directing resources at either reducing cancer incidence or accident rates, you might naturally assume that reducing the incidence of cancer would have the biggest effect on life expectancy overall.

However, it is also important to consider the respective impact on population dynamics and the age distribution. Arnold and her co-authors modelled a situation where both accident and cancer incidence were supressed at a level which produced similar life expectancy gains. But while the life expectancy gains were similar, the impact on the age distribution was markedly different. Cancer affects mainly older people. When you supress cancer you save a lot of lives but at older ages. When you supress accidents you save a lot of people at a younger average age – around 25-30 years old. These people, who now do not die, may well have children. Thus, in the long term, there is a very different impact on the age structure of the population and on the old age dependency ratio.

The implications of these findings are clearly important for policymakers, but they are also relevant for commercial organizations. Many companies have business models that depend on successfully serving the needs of certain sections of the population. As a result they devote considerable resources to forecasting demographic trends. Now they have another tool that they can use. Because the model that Arnold and her co-authors have developed allows organizations to more accurately assess the impact of events that affect life expectancy – a new policy initiative, a new pharmaceutical product or service – on the composition of the age structure of future consumer populations.

Related research paper:

Alexandre Boumezoued, Héloïse Labit Hardy, Nicole El Karoui, Séverine Arnold (2018). Cause-of-death mortality: What can be learned from population dynamics? Insurance: Mathematics and Economics 78: 301-315.

Featured image by © Jakub Jirsák ID 84747574