September 2010 Past Issues | Advertise |
Return to main page

Cohort: Meaning and Insights

Print Print this Article | Send to Colleague

By Abinash Churoria

Not many people realize that a year in which a person is born can be a critical factor in determining how long they will live. Welcome to the fascinating world of cohort research in particular, and mortality research in general.
In actuarial science we come across cohort a number of times, but seldom do we explore deeper. Its dictionary meaning is a group of people who share a common feature or aspect of behavior. In mortality research what we consider primarily is a cohort of people born in the same time period, say in the same decade—more commonly known as "year of birth cohort." Though a "cohort effect" can be observed in any population, the cohort effect in the UK’s population is quite pronounced and well researched.

The Continuous Mortality Investigation Bureau (CMI) is an organization of the Actuarial Profession whose mandate is research into the UK’s mortality and morbidity experience. It is funded by its members, the majority of whom also submit data. They also receive regular reports. CMI has sponsored research into the detailed examination of the cohort effect, the results of which were published in 2002 in Working paper 1.

In the context of the UK, cohort effect denotes a phenomenon whereby people born between 1925 and 1945 (centered on the generation born in 1926) have experienced more rapid improvements in mortality than people on either side of the period. A graphical representation of the cohort effect may give a more insightful view.

Figure 1. Mortality improvement by age and decade for males in the population of England and Wales. Data source: Office of National Statistics (ONS), 2001.

The data used to produce figure 1 are given in table 1. The age groups showing the fastest improvement are highlighted in bold type and can be seen to move diagonally, indicating that the same birth cohort has consistently experienced the most rapid mortality improvement.

The equivalent information for females is given in table 2.  The figures in table 2 indicate that exactly the same "cohort effect" has applied for females as males. This is notable given that the relative importance of different causes of death varies by gender.

Table 1
Percentage mortality improvement by age and decade for males in the population of England and Wales. Data source: ONS, 2001.

 Age group
   1960s  1970s  1980s  1990s
 25-29 1.5 0.1 0.4  -0.9
 30-34 1.7 1.4  -0.6  -0.8
 35-39 1.7 1.0 0.3 0.9
 40-44 0.1
2.1 2.1
 45-49 -0.2
 50-54 0.2
 55-59 1.0
 60-64 1.0
 65-69 0.1
 70-74 0.1
 75-79 0.7

Table 2
Percentage mortality improvement by age and decade for females in the population of England and Wales. Data source: ONS.

 Age group
   1960s  1970s  1980s  1990s
 25-29 1.6
 2.6  0.4
 30-34  2.5  1.2  0.8  0.6
 35-39  1.7  1.5  1.4  0.7
 40-44  0.5  2.0  1.7  0.3
 45-49  0.4  1.8  2.2  1.2
 50-54  0.0  0.6  2.7  1.4
 55-59  0.3  0.3  2.0  2.0
 60-64  1.1  0.3  0.9  2.7
 65-69  1.2  0.9  0.6  2.4
 70-74  1.4  1.3  1.0  1.2
 75-79  1.6  1.2  1.5  1.0

What was so unique about people born between 1925 and 1945 that their mortality rates were markedly lower from those of people born either side of that period? Experts highlighted a combination of primary factors that led to this cohort effect:

a) World War II: People born in the 1930s and early 1940s did not fight in a major global conflict, as earlier generations did in the form of World War II. So the cohort effect can be attributed to more adverse effects experienced by those previous generations rather than beneficial effects experienced by the later generations (born in 1930–1945).

b) Diet: The diet in postwar Britain had health benefits for children growing up in that period. Food rationing lasted until the 1950s, but the average consumption of fresh vegetables, milk, potatoes, bread, fish, etc., was higher during the postwar years. Conversely, the consumption of meat and cheese was lower during the same period.

c) Welfare state: the period after 1940 was a time of great social change in the UK. For instance, the right to secondary education was made universal in 1942 and the National Health Service was launched in 1947. Hence the social environment in the 1940s was very different to that of previous decades.

d) Smoking histories: the smoking histories of various generations are very different. During World War II cigarettes were distributed free to those on active service, and many of those born in the 1920s and earlier were probably in the forces and smoking profusely. But it was after the war that tobacco’s hazardous health effects began to be investigated in earnest. So by the time those born in the 1930s reached adulthood the link between smoking and adverse health effects was becomingly increasingly well-known. Consequently, cigarette consumption slowly began to fall. Also, the so-called Doctors Study which investigated the impact of smoking patterns on mortality differentials between smokers and non-smokers was initiated in 1951.

e) Historic birth rates: last but not least is the fact that the "high improvement birth cohort" coincided with a trough in birth rates relative to the period immediately before and after it. A graphical representation will clarify the point.


Figure 2. Live births in England & Wales – 1900 to 1970. Data source:

Birth rates increased by 31 percent between 1941 and 1945, which may be quite relevant. One possible consequence of dramatic change in the rates is that the "average" child is likely to be quite different from children born immediately before and after that period. Further evidence of this can be found if we plot the birth rates over this period by socio-economic class. Amid declining birth rates, if we find that the rates of the affluent class have improved vis-à-vis the not-so-affluent class, the population’s socio-economic mix changes.

Once we are familiar with the causes, we need to ask, how does it affect various actuarial tasks? The mortality projections released by CMI in its report CMIR 17—more popularly known as "the 92 series"—did not incorporate the impact of these birth cohorts. So in 2002 the CMI produced the Interim Cohort Projections, a revised set applicable to the 92 series:

a) Long cohort: under this set of improvements it was presumed that the cohort effects would tail off by the year 2040.
b) Medium cohort: under this set it was presumed the effects would tail off by 2020.
c) Short cohort: under this set it was presumed the effects will tail off by 2010.

The world of mortality analysis is an intriguing field of research with the study of cohorts being just a small part. Predicting death is not an easy task, but at least small efforts can be made to understand the way human lives have evolved and are evolving over time. As singer Kanye West said: "Nothing in life is promised except death."


CMI. Working paper 1
Richard Willets, FFA. ‘Mortality in the next millennium.’ Presented to the Staple Inn Actuarial Society, December 7, 1999.

Abinash Churoria is a student member of the Institute of Actuaries of India, and the Faculty and Institute of Actuaries.

360 Albert Street, Suite 1740, Ottawa, ON K1R 7X7
Tel.: 613.236.8196 Fax: 613.233.4552