Statistical Insights: Large inequalities in longevity by gender and education in OECD countries
While differences in average longevity, or life expectancy, between countries are well-documented, inequalities in longevity within countries are less well-understood and are not fully comparable beyond a handful of European countries. A recent OECD working paper (Murtin et al., 2017) fills this gap by analysing inequalities in longevity by education and gender in 23 OECD countries in 2011.
Measures of inequalities in longevity show that, on average, the gap in life expectancy between high and low-educated people is equal to 8 years for men and 5 years for women at the age of 25 years; and 3.5 years for men and 2.5 years for women at the age of 65. Cardio-vascular diseases, the primary cause of death for the over 65s, are the primary cause of mortality inequality between the high and low-education elderly.
Figure 1 shows the longevity gaps between high and low-educated people at the age of 25 and 65. At age 25, life expectancy is 48.9 years for men with low education, 52.6 years for those with medium education and 56.6 years for those with high education. The corresponding figures for women are 55.5 years, 58.3 years and 60.1 years respectively.
Longevity gaps differ markedly across countries. High-educated 25 year-old men for example can expect to live more than 11 years longer than their low-educated counterparts in Latvia, Poland, the Czech Republic and Hungary, while the gap is less than 5 years in Portugal, Turkey, Italy, New Zealand and Mexico. In the case of women, inequalities in life expectancy are relatively small in Austria, Israel, Portugal and Italy, but amount to over 6 years in Latvia, Poland, Belgium and Chile.
Large inequalities in longevity by education persist even at older ages. At 65 years, life expectancy for men, on average in the OECD, is 15.8 years for those with low education, 17.1 for those with medium education and 19.2 years for those with high education. The corresponding figures for women are 19.6, 20.8 and 21.9 years. In relative terms, i.e. expressed as a share of the remaining lifespan, gaps in longevity are larger at 65 than at 25.
While differences in average life span (i.e. longevity or life expectancy) between groups of education and gender are large, this masks wider differences in life span within groups when other factors, such as genetics and exposure to risk factors are taken into account. Indeed, combined, education and gender, only account for around 10% of the total variation in lifespan.
Breaking down mortality rates (measured as the probability of death in a given year) of people aged between 65 and 89 years by causes of death (circulatory causes such as heart failure, neoplasms or cancer, external causes such as accidents, and other causes) reveals that circulatory problems are the leading cause of death for both gender and education groups (Figure 2). Indeed they account for about 40% of total mortality, with neoplasms and other causes of death accounting for between 25% and 30%. Circulatory problems are slightly more prevalent among the low-educated, for both men and women.
Focusing on low-educated older men, circulatory problems are the most frequent cause of death in high-mortality countries such as Latvia, the Czech Republic, Poland and Hungary, where they account for around half of all deaths, as compared to around one third of deaths in Canada (28%), the United Kingdom (30%), Norway (37%) and Turkey (31%). Conversely, other causes of death are relatively more prevalent in low-mortality countries.
Circulatory problems are also the main factor explaining the mortality gap between education groups at older age. For elderly people, circulatory diseases contribute to 41% of the difference in mortality rates between low and high-educated men and 49% between low and high educated women.
Addressing the risk factors underlying circulatory diseases, in particular smoking, seems as an efficient way of reducing both average mortality rates and inequalities in longevity across education groups. According to Mackenbach (2016), smoking accounts for up to half of the observed inequalities in mortality rates in some European countries; also, while its contribution to inequalities in longevity has decreased in most countries for men, it has increased among women.
The measure explained
Longevity is statistically defined as the average number of years remaining at a given age. It is calculated as the mean length of life of a hypothetical cohort assumed to be exposed since starting age until death of all their members to the mortality rates observed at a given year. Mortality rate is a measure of the number of deaths in a particular population, scaled to the size of that population, per unit of time.
Estimates of life expectancy by education are drawn from data compiled in different ways in different countries. Two main approaches (study design) can be distinguished:
- A “cross-sectional’ (unlinked) design implies that information on the socio-economic characteristic of the deceased is drawn directly from death certificates, as reported by relatives or public officials, while population numbers for the same population categories (the denominator of mortality rates) are drawn from the most recent population censuses. An obvious drawback of this “unlinked” design, which is still used by most countries reviewed in the accompanying OECD working paper, is that it can only be implemented when death certificates include information on the occupation and education of the deceased. In addition, even when this information is included in death certificates, it may be affected by (large) recording errors;
- A “linked design” implies that socio-economic information on the deceased is retrieved by individual data linkage to the most recent population census or administrative register records. While both types of data are used in this study the “linked approach” is generally associated with higher quality data. Beyond these differences, a number of data treatments are implemented to correct for statistical biases and anomalies that may arise when calculating mortality rates based on a small number of deceased with specific age, gender and education characteristics.
The measure of inequalities in longevity described above have two specific features that may affect cross-country comparisons: first, life expectancy is disproportionately affected by mortality rates at very young ages compared to mortality rates at older ages; second, measures of life expectancy by education are also affected by the fact that distributions of education vary across countries and time. Alternative measures of inequalities in longevity have been examined, and they all show very large correlations across countries. In other terms, accounting for differences in the size of the various educational groups or using average mortality rates rather than life expectancy measures, does not change the assessment of countries’ rankings significantly. There are also significant cross-country differences between these measures of inequality in longevity and more traditional measures of inequality in, for example, income: longevity inequality are lowest in Italy, where income inequality is relatively high, and highest in many Eastern European countries, where income inequality is relatively low.
Where to find the underlying data
The underlying data can be found online at the following address: www.oecd.org/std/Inequalities-in-longevity-by-education-in-OECD-countries.xlsx
Mackenbach, J.P. (2016), Health Inequalities in Europe, Erasmus University Publishing, Rotterdam
Murtin, F., Mackenbach, J.P., Jasilionis, D. and M. Mira d’Ercole (2017), “Inequalities in Longevity by Education in OECD Countries: Insights from New OECD Estimates”, OECD Statistics Working Papers, 2017/2, OECD Publishing, Paris .
In the 18th century, the Church of Scotland had a problem – widows. Whenever a married minister died, his salary was transferred to his successor. If the dead man was married, his widow and children could find themselves without either income or home. The only option for some was the poorhouse – an unedifying spectacle for a church.
Two churchmen, Dr. Robert Wallace and Alexander Webster, decided something needed to be done. They wanted to create a fund that would provide ministers’ widows with a pension. But how to design it? Rather than turning to prayer, the two sought a solution in another of their great passions – mathematics.
As Yuval Noah Harari explains in his excellent Sapiens, the two churchmen exploited the emerging field of probability and, in particular, the Law of Big Numbers. This stated, in effect, that while “it might be difficult to predict with certainty a single event … it was possible to predict with great accuracy the average outcome of many similar events”. So, while it was impossible to say how many years an individual 60-year-old minister might live, one could say how much longer the average 60-year-old man would live. All that was needed was good data.
Here the two ministers were in luck. At the end of the 17th century, the great astronomer Sir Edmund Halley – better known today as a comet spotter – had studied records for births and deaths in the German city of Breslau. From these, he produced some of the world’s first life tables, which gave precise odds for the probability that a person of a certain age would die in a certain year.
The two preachers used this data to design their fund. As Dr. Harari explains, the fund proposed a range of pension plans: For example, any minister who paid in £2.12s.2d (or about £2.61) a year would guarantee an annual pension for his widow of £10 – more than enough to keep her out of the poorhouse.
The fund was an enormous success, not just because of the peace of mind it brought to ministers’ families but also because its careful design meant it was financially sound. Wallace and Webster had predicted that 20 years after its establishment, the fund would be worth £58,348. They were wrong: In fact it was worth, £58,347 – just £1 short of their forecast.
If only things were as straightforward today. The church fund existed at a time when – outside periods of disease and war – life expectancies changed little. That’s not the case now. In a typical wealthy country today, the life expectancy of a 65-year-old is rising by nearly two months every year.
Rising life expectancies are, of course, a good thing, but they do create problems for pension funds, most notably “longevity risk” – in other words, pension funds aren’t always taking sufficient account of just how much longer people are living. That’s due in part to the fact that they may be relying on outdated data that doesn’t improvements in life expectancy. According to a new OECD report, getting these estimates wrong can be costly: “Each additional year of life expectancy not provisioned for can be expected to add around 3-5% to current liabilities.”
Rising longevity is posing other problems for pensions, according to the OECD Pensions Outlook 2014, released today. In many countries, the rising share of retirees in the population will leave more people dependent on a shrinking share of workers. This imbalance will become much more evident as growing numbers of post-war baby-boomers reach retirement age.
There are other uncertainties, too, notably what Larry Summers (pdf) calls “secular stagnation” – a drawn-out period of economic sluggishness “characterized by low returns, low interest rates, and low growth”, says the Outlook. Meeting pension commitments against that backdrop could become quite a challenge.
How are governments responding? The Outlook reports that the pace of reforms has speeded up. Priorities have included raising the retirement age and linking benefits to expected rises in longevity as well as steps to strengthen the funding of private pensions. And to respond to rising public concern over private pension providers, there’s a growing focus on how the sector is regulated. Still, there’s no doubt that much more will need to be done if today’s future pensioners are to enjoy the peace of mind of those 18th century Scottish widows.
OECD Pensions Outlook 2014 (OECD, 2014)
Mortality Assumptions and Longevity Risk (OECD, 2014)
OECD work on insurance and pensions
“Don’t Stop Working!” Not our advice, but the headline on a Slate article about one of the world’s longest-running sociological studies. Back in the 1920s, the American psychologist Lewis Terman gathered together 1500 exceptionally bright boys and girls and began a study of their lives that would continue for eight decades. Today, only a few of Terman’s “Termites” are still alive, and final conclusions from the study are being published.
One aim of the research was to find out what makes some people live longer than others. The findings are interesting: For instance, the death of a parent in childhood had relatively little impact on longevity, but parental divorce did. Adults who were gloomy and neurotic as children also tended to die relatively young. But so did those who had been extra cheerful, perhaps because of a devil-may-care attitude towards smoking and drinking in later life. The real winners in the longevity stakes were the conscientious kids, those who as adults maintained “a fairly high level of physical activity, a habit of giving back to the community, a thriving and long-running career, and a healthy marriage and family life”, as The Wall Street Journal puts it. Hence Slate’s advice to keep on working.
That may also be music to the ears of governments, which increasingly want to see people working later in life. As the latest edition of OECD Pensions at a Glance reports, around half of OECD countries have already started, or are planning to start, raising “pensionable ages” – the age at which people qualify for a full pension. By 2050, the average in OECD countries will reach just under 65 for both sexes – that’s nearly 2½ years above the current age for men and 4 years for women.
A key reason for this move lies in the fact that we’re living longer. As a result, most of us will be living off pensions for much longer than our grandparents did. In 1958, a man who reached pensionable age could look forward to living for around another 13 years; by 2050, that number is forecast to rise to just over 20 years. The figures for women are perhaps even more striking: 17 years in 1958, but 24½ years – almost a quarter of a century – in 2050.
Paying for all this risks being a major strain on the taxpayers of the future, especially as there will be relatively fewer of them than today: In most OECD countries, declining birth rates mean that non-working over-65s will account for an ever larger slice of the population. In 2000, about a third (33%) of people in OECD countries were aged over 65; by 2050, that number is forecast to exceed 41%.
So, in all probability we’ll all have to go on working a bit longer. Not everyone’s happy with that: In France, unions argue the burden will fall unfairly on blue-collar workers. But it’s interesting to note that the higher retirement ages of the future will in some ways take us back to where we were: Over the second half of the 20th century, the average pensionable age in OECD countries fell by two years, before beginning to rise again in the 1990s. If the forecasts are accurate, by the time we reach 2050 it will be only about 3 months above what it was in 1948, or 64.6 years.