J. N. Mojekwu
Department of Actuarial Science and Insurance, University of Lagos, Lagos
J .S. Adeyele
Joseph Ayo Babalola University, Ikeji-Arakeji
The study investigates mortality patterns of civil servants and its effects on pension reform scheme in Nigeria. Attempts were made to examine the relationship between income and retirement decisions as well as the relationship between income and mortality patterns. Differentials due to demographic factors such as sex, age, and life expectancy were considered. A sample size of 796 federal civil servants who retired between 1974 and 2008 in Lagos metropolis was used for this purpose. Correlation and Chi-square distributions were used to test the hypotheses at 0.05 significant level. The results seem to suggest that there is strong relationship between income- retirement decision and income- mortality of civil servants. Also, the study revealed that male mortality is lower than female mortality amongst the retiring civil servants and hence could not support the claim that life expectancy has no relationship with the time of retirement. Based on these findings ,the  present pension reform is too new for its long term implications to be evident.

Keywords: Mortality patterns, civil servants, pension reform schemes, retirement income decision

 Pension system in Nigeria has not been able to contain the retirement cost over three decades. The new Pension Reform Act 2004  appears to many pensionable workers as a viable way to resolve pension crisis in Nigeria. Following the implementation of this reform, employees are requested to open a Retirement Saving Account with the Pension Fund Administrators upon which   balances standing to their credits at age 50 will be used for the following benefits:

  1.  Programmed monthly or quarterly withdrawal calculated on the basis of an expected lifespan; and
  2. Annuity for life purchased from a life insurance licensed by the National Insurance Commission (NAICOM) with monthly or quarterly payments.

In light of this, the reform in pension system now creates new challenges to life insurance companies. One of the challenges mentioned by the practitioners is the designing of a broad range of the annuity products. Another prominent challenge usually mentioned by the practitioners is pricing the designed annuity products to match the benefit of retirees. However, annuity cannot be equitably priced without understanding the mortality pattern active Civil Servants are likely to experience.

Hence, there is a great need to investigate mortality patterns of civil servants and its effects on pension reform scheme in Nigeria. Also, the examination of implications of the present schemes on stability of tenure is necessary in order to advise the pensioners as well as the pension operators for effective economic growth. The relationship between retirement decision   and mortality patterns is a vital indicator to assess the contribution of the new pension scheme to economic growth. Also, the study of relationship between retirement income and retirement decisions can assist in measuring the performance of the scheme. Differentials due to demographic factors such as sex, age, and life expectancy were considered in the study to buttress the factors actually contributing to the success and failure of the scheme.

Objective of the study
The study attempts to examine the relationship between  retirement mortality factors and present pension reform schemes. More specifically, the study attempts to

  1.  examine whether income per person influences the person’s decision to retire and 
  2.  determine whether there is mortality differentials in retiring civil servants due to sex.

Research hypotheses
To critically address the objectives of the study, the following hypotheses were formulated.
(i)         Income per person does not relate to retirement decision of an
(ii)        There is no significant difference between male and female mortality.

Willet (2004) and Norbreg (2008) revealed that huge increases in life expectancy in advanced countries which led to longevity risk is partly responsible for  pension crisi, Blake et al (2006) revealed that the possible consequences of longevity risk came to public prominence in December 2000, when the world’s oldest life office, the Equitable Life Assurance Society (ELAS) was forced to close to new business – life annuity. Weil and Kalami-Ozcan (2005) view this rise in retirement as one of the most dramatic economic changes that have taken place in the last hundred years. In order to guard against longevity risk in the present reform, there are variables of interest to be used in charging annuity products. Modugno (2003) Wadron (2001) asserted that the challenges facing the operators of pension schemes are the designing and pricing of annuity products.

Some of the important variables used for charging annuity products are: age, gender, income and occupation. The civil servants chosen for this study are presumed to experience less risk compared with other professionals like Pilot or mining workers. Based on the chosen workers, the link of income, gender and age to mortality are discussed as follows. Institute of  Actuaries and Faculty of Actuaries (1999) noted that the mortality pattern of early retirees and those who retired later is not significantly different.
 Lentz et al (19998),  Deaton and Paxton (1999) and Brown and McDaid (2003) found that wealthier people have lower mortality rate compared with poorer people regardless of sex, age, race or marital status. Bozzoli (1985) in Brown and McDaid (2008) noted the inverse relation between income and mortality. That is, as income increases mortality reduces.  Since income variable is likely to be taken into account in determination of annuity price, the study evaluates the level of relationship between income and mortality. Table 2 reveals that pensioners with the highest mean income have the lowest mortality rate in all age specific death rates (ASDR).

Gender is also considered to be a very important variable in the pricing of any life policy. Over several years, insurance companies have relied on the assumption that the male has shorter lifespan compared the with female. Winklevoss (1976) commented that it was not entirely clear why females tend to have lower mortality rates than the males at every age. Brown and McDaid (2008) considered socioeconomic groups, based on occupation and found that occupation was related to excess male mortality. Deaton and Paxton (1999) revealed that income has positive effect on mortality among young men, but the effect was large and negative for young women.

This study considers mortality rates between males and females. Table 3 shows that males’ mortality is lower than females’ at every stage of retirement. This result is at variance with the general beliefs that female has lower mortality than male.

More so, chi-square (X2) distribution and correlation were both used to test the result presented in table 3 at 0.05 significant level, with 2 and 1 degrees of freedom respectively. Opposing results were obtained but the latter result was used for conclusion by the researchers. The essence of this test is to challenge the assumption that more males die than females. This study shows that if both sexes are exposed to the same risk, females have no advantage over males. Consequently, the researchers concluded on the basis of table 3 that males have lower mortality rates than females at every stage of retirement if they are both exposed to similar risks.

Age at retirement otherwise called retirement decision is often another variable of interest in all the designs of life policies. As a result, many studies have attempted to link retirement age with state of health or life expectancy, which in turn, used to determine the price of life policies. Average life expectancy of Nigerians as reported by World Health Organization (2009) is 49 years. This study directly examined the link between early retirement and normal retirement of those still living and the deaths. The average ages of early retirees and normal retirees at retirement are 41.61 and 57.89 respectively.
Cheng (2007) claims that early retirement leads to greater life expectancy and vise versa. This claim was refuted by Patersons (2005) who states that early retirement does not lead to longer life. Paterson supported his claim that industrial workers who retired at 60 or even 55 do not live longer than those who worked until 65.  Bamia et al (2008) attempt to throw light on this controversy that link between life expectancy of early retirement and normal retirement has not been established. That is, no significant difference between them. Institute of actuaries and Faculties of Actuaries (1999) noted  that mortality experience of early retirees are expected to improve relative to that of normal retirement.

The three variables of income, gender and age discussed so far happen to be poorly understood by many life companies. The lack of their proper understanding has led to pension crisis in many advanced countries. In most life policy designs, life expectancy in advanced countries was underestimated. This led to inability of life companies and government to meet up with pension obligations as revealed by Blake et al (2004). At present, the 49 years of life expectancy as reported in World Health Organization (2009) appears to be very low, but this will improve over time relative to the level of medical advancement in the country.

Data collection
The population for this study is made up of all the Federal Civil Servants who registered with Pension Fund Administrators in Lagos metropolis and covers the 23 licensed Pension Fund Administrators (PFA) in Nigeria.  One of these PFAs was discretionally selected as it reflects a good representation of the characteristics of civil servants than the remaining 22 PFAs. 796 pensioners were extracted from the selected Pension Fund Administrators’ records with the following distributions which yielded 598 males alive, 27 male dead, 166 females alive and 9 females dead. The data used for this study was extracted from the selected PFAs’ records. This covers pensioners who retired between 1974 and 2008 who are currently receiving pension benefits.

This study used statistical tools such as chi-square and correlation to test the formulated hypotheses. All hypotheses were tested at 5 percent significant level.

The total sample size of 796 Civil Servants who retired between 1974 and 2008 is summarized in table 1. The total income at each age group is divided by the total number of Civil Servants in those age groups to get mean income per person. Similarly, the total incomes at sub-periods were divided by the total number of Civil Servants in those periods to get the overall mean income per person. The mean income per person between age groups was compared with the mid-ages. All the statistical tools such as correlation which provide a measure of the strength of association between two variables; one dependent, the other independent and Chi-Square are displayed below

                r = ……………………                (1.1)

where   . A perfect relationship exists, where r =  1. While
 the other formula used for testing the hypotheses is also shown below
                           ……………………… …….         (1.2)
   where fo = observed frequency in a particular cell of contingency table
             fe    = expected frequency in a particular cell if null hypothesis  is true.
Information in table 4 was derived from table 1 to obtain the test statistic given by             = 7.4229

At 5 degree of freedom, the critical t is 2.571. Since 7.4229, is greater than 2.571, then the numerical evidence is strong enough to reject the H0.

Chi-square and correlation were used to examine relationship between gender and mortality. Table 5a indicates that the calculated x2   value is 0.384. At 5 percent confident level, with 2 degrees of freedom, the critical x2 value is 5.99. Since the calculated x2   is less than the critical value, it appears that the null hypothesis should be upheld. However, final decision would be arrived at by the use of correlation.
In table 5b,   r =0.9405, r2 =   0.8848   S = significant at 0.05; df = 1; Critical t = 6.314
H0:  = 0 versus  H1: 0




0.9406 – 0
1 – 0.8848

Using the information in table 5b, the test statistic is obtained by:

At 0.05 significant level with 1degree of freedom, the calculated statistics is 8.1644. Since 8.1644 is greater than 6.314 we reject the null hypothesis and conclude that there is significant relationship between sex and mortality and that male mortality is less than the female mortality at every stage of retirement (see table 3). This result contradicts the findings by Winklevoss (1976) that females have lower mortality rates than males at every age.

This study views the implications of the new pension reform schemes in Nigeria from two perspectives. The first is concerned with stability of tenure while the second deals with the challenges of the new schemes to annuitants (retirees) and the life companies. To allow for effective discussion on stability of tenure, the study attempts to answer the first research question stated.

Data presented in table 5 shows that retirement decision (age at retirement) is based on income. In section 1.3 of pension reform Act 2004, employee is permitted to maintain his RSA if he wishes to transfer his service from the present employer to the other. This implies that employee is entitled to100 percent of the amounts contributed to his RSA irrespective of the number of years of service in that employment. Since income is linked with retirement decisions, this present scheme is likely to encourage high turnover of employees in organizations that cannot pay competitive salary, thereby affecting productivity. Employee turnover is a costly  especially in lower paying job roles, for which the employee turnover rate is highest. Although, this study did not directly examine other factors that lead to retirement decisions, many factors play a role in the employee turnover rate of any company, and these can stem from both the employer and the employees. Wages, company benefits, employee attendance, and job performance are all factors that play a significant role in employee turnover.  Since one of the research findings reveal that income is strongly related to retirement decision, the ability of employer to structure schemes that motivate desirable behaviour such as  longevity and loyalty of employee is much weaker in the present scheme.
As a rule, the present scheme requires all employees to open RSAs with PFAs of their choices. These accounts as a matter of interest introduce two elements of risk to pensioner’s income – namely, investment risk and administrative charges risk, and these may lead to significantly different outcomes from those available in the old pay-as-go (PAYGO) basis in public schemes. The move from DB to DC also implies that the longevity risk will be borne by pensioners rather than the employers and no redistribution is possible.

To permit redistribution in annuity markets, section 4 of pension reform Act 2004  provides that a holder of a RSA, upon retirement or attaining the age of 50, whichever is later, shall utilize the balance standing to the credit of his RSA for retirement benefits. In addition to these benefits, subsection 3 of section 9 of Pension Reform Act 2004 also provides that  employers should maintain life insurance policy in favour of their employees for a minimum of three times the annual total emolument of the employee.
Following the implementation of this Act, life companies have new line of businesses but faced with two major challenges: how to design a broad range of products and price them accordingly.

In the traditional life insurance policies, longer life may not mean a crisis, but at least, in financial terms, it is a challenge in the design of annuity prices. This is the risk to which a pension fund or life insurance company could be exposed as a result of higher-than-expected payout ratios. The compulsory purchase of annuity products by the retired employees under the present schemes might not lead to quiet revolution. Quiet revolution in pension policy as used here means that the financial costs of longer lives will be shared among the policyholders subject to a rule rather than between generations as obtained in the old PAYGO in public schemes.

Average life expectancy of Nigerians as reported by World Health Organization (2009) is 49 years. The average ages of early retirees and normal retirees at retirement are 41.61 and 57.89 respectively. Also, average age at death of early retirees and normal retirees are 53.56 and 63.94 respectively. From this figures, those who retired earlier will have at least 11.95 (53.56-41.61) more years to live while those who retired later have about 6.05 years (63.94-57.89) to live. Based on this study, both early and later retirees’ life expectancies are greater than the one reported by the WHO. If annuity charges were to be based on life expectancy at birth, later retirees will be charged higher than the early retirees because life expectancy of the former is higher than the later. On the other hand, if charges are based on life expectancy at retirement, early retirees will be charged higher than the later retirees because the later have fewer years to enjoy annuity benefit than the former.
The above finding support Cheng (2007).  Although by the use of Chi-square X2 at 0.05 significant level, with 1 degree of freedom, the result was not significant. That is to say there is no significant difference in life expectancy of people who retire earlier and those who retire later. If life expectancy were to be based at birth, Cheng (2007) that says early retirement leads to greater life expectancy will be correct. It is expected that mortality experience of early retirees would improve relative to that of normal retirement.
 Another implication of the above findings to life companies is that, if future mortality improves relative to expectation, the life insurance liabilities decrease because benefits payments will be later than expected initially. However, if the mortality deteriorates, the situation is reverse: life insurance has losses and annuity writers have gains.

Based on the available information in this study, the following findings were made:

  1. Income and retirement decision play an important role in stability of tenure.
  2. Those who retired earlier have more years to live than those who retired  at normal retirement.
  3. Given the same risk exposures, there is no significant difference in male and female mortality.
  4. Irrespective of retirement age or gender, all retirees within homogenous group or the same profession have relatively the same mortality rate.

 If life companies are to achieve complementary role of this pension reform, they must be willing to design varieties of products to suit the needs of individual retires. Consequently, annuity underwriters will need to rely on professionals like actuaries in order to provide viable and acceptable service to the annuitants. The role expected from actuaries in this case would be the development of workable models in Nigerian environment since the existing life tables used for charging life policies may not be adequate. Hence, the need to develop empirical basis for determining mortality factors in Nigeria. Once this is achieved, appropriate modification must be effected on the existing life tables.

The present pension reform is still very new for its long term implications to be evident. It appears that mortality deterioration will lead to more risk classification in annuity markets. As a result, if professionals like actuaries are not encouraged to assist life companies design appropriate models for pricing annuity products in good time, the present pension systems is waiting for another explosive crisis.
Finally, considering the strong association between income and retirement decision, there is a need to redesign pension schemes in such away it can reduce job turnover among employees as well as increasing productivity in organization. Employers must take a deep interest in their employee turnover rate because it is a costly part of doing business.

Bamia, C., Trichopoulou, A. and D. Trichopoulou (2008). Age at Retirement and Mortality in General Population Sample. Journal of Epidemiology (http//
Blake, D., Cairns, J.A.G and Dowd, K (2006). “Living With Mortality: Longevity Bonds And Other Mortality-Linked Securities” British Actuarial Journal, vol.(10), part IV, pp.153-228.
Brown, R. L. and McDaid, J. (2003). Factors affecting retirement mortality North America Actuarial Journal. Vol.7(2); pp.24-43 (http// farm.html).
Brown, R. L. and McDaid, J. (2008). Factors affecting retirement mortality Phase I ( mortality-farm-introduction.aspx).
Cheng, E. (2007). Retirement vs Life Expectancy (asssed at: http//: new2the expectancy.html.)
Deaton, A. and Paxton, C. (1999). Mortality, Income, and Income inequality over time in Britain and United States
(  )

Institute of Actuaries and Faculty of Actuaries (1999). “The Mortality of   Pensioners In Insured Group Pension Schemes 1991-94” Continuous Mortality Investigation (CMI) Report No.16.
Kalemi-Ozcan, S. and Weil, D.N (2001) Mortality Change, uncertainty Effect and            
                    Retirement. New York: John Willey and Sons

Lantz, P. M. House, J. S. Lepki, J. M. and D. R. Williams (1998). Socioeconomic factors, health behaviours, and  mortality results    from a Nationally Representative perspective   study  of U.S Adult. Journal of America Medical Association. Pp 279;1701-1708.


Modugno, V.(2003). Factors affecting retirement mortality (FARM)-phase II. (

Norbreg, R (2008). the pension crisis: its causes, possible remedies, and the role of the regulators.

O’brien, C. D. (2002), “Guarantee Annuity Options: Five Issues for Resolution” British Actuarial Journal, Vol.8(37), Part III,  pp. 593-629.
Pattersons Michael (2005).Early Retirement does not lead to Longer live.  News Research British Medical Journal. (

Waldron, H.(2001). Link Between Early Retirement and Mortality. Division of Economic Research. Social Administrators’ Office of Actuary

Willets, R. C. (2004). “Longevity in the 21st century” British Actuarial journal, Vol.10, part IV, pp.637-790.

Winklevoss, H. E. (1976). Pension mathematics with numerical illustrations. Irwin Inc: Philadelphia.
Worl Health Organisation (2009) Life expectancy in Nigeria


Table 2: Pensioners, 1974-2008: Exposed to risk, death, crude death rate and average pensions.

Age Group                                       Exposed to Risk               


(Nearest ages)    Lives        Amounts(pa)   Average income(Npa)

Lives  ASDR  Amounts(pa)  Average income(Npa)

40-44                   5              489,852.02                  97,970.40


45-49                  83            10,717,101.44             129,121.70

2          0.0241     610,283.44              305,141.72

50-54                  98            24,562,750.31            250,640.31

10       0.092        3,912,366.32           391,236.63

55-59                  141          57,522,262.28            407,959.31

4         0.035        1,147,275.41           286,818.85

60-64                  222          160,131,230.30          721,311.85

13       0.054        10,025,104.88         771,161.91

65-69                  155          79,047,760.4              509,985.55

3         0.0194      3,085,393.63          1,028,464.54

70-74                  71            25,433,004.09           358,211.33

4         0.042        1,518,906.61           379,726.65

75-79                  17            3,035,844.74             178,579.10


80-84                  2              932,308.49                 466,154.25


85-89                  1              503,539.68                 503,539.68


95-99                  1              91,987.01                   91,987.01


40-99     796 (0.955)*     362,467,640.8            455,361.36

  36  CDR 0.045   20,299,330.29   563,870.29

Source: Extracted from Pension Funds Administrators’ Records *Survival rate.

Table 3:           Pensioners Exposed to Risk, Crude Death Rate and Deaths between 1974-2008

Retirees in pension payment   Exposed to Risk         Deaths        Age Specific  Death Rate                    

  Males :  Withdrawals (30-44)           169                         5                                   0.0296
            Early Retirement(45-49)          76                         7                                   0.0921
           Normal Retirement (50-64)     376                       15                                  0.0399
     Females:  Withdrawals (30-44)       41                         2                                  0.0488
                   Early Retirement(45-49)   34                         3                                   0.0882
               Normal Retirement (50-64) 100                         4                                   0.04

Total                                                   796                       36             CDR    =      0.0452

Source: Extracted from Pension Funds Administrators’ Record

   Table 4a: Relationship between income per person and retirement decision






Calculated statistic







r = 0.9575

r2 = 0.9168

DF = 5

N = 7

t = 2.571


Critical t = 2.571, S = Significant at 0.05
             Table 5a: Link between sex and mortality.

Male (x)

Female (y)










Df = 1







          r =0.9405, r2 = 0.884

           Table 5b x2 : Link between sex and mortality.


Withdrawal Early Retirement  Normal retirement




3(3.76)           9(8.45)           4(3.76) 
5(4.24)           9(9.45)           4(4.24)             




8                     18                    8



           Expected frequencies are in bracket; NS = significant at 0.05; DF = 2; critical x2 = 5.77.