The Costs of Improvements in Romanian Life Expectancy
The background of this work is the implementation in Romania - as of the year 2007 - of the private pension funds system. This system has two pillars: second pillar (mandatory for workers under age 35 and optional for workers between age 35 and 45) and third pillar (facultative pensions). In the second pillar, workers between age 35 and 45 have had the option till January 2008 to join the mandatory layer (for them it becomes mandatory from the day they choose to join a mandatory pension fund). Workers can join the third pillar whenever they want. According to figures published on the National House of Pensions and Other Social Insurance Rights more than 1.5 million workers between age 35 and 45 have chosen to join the mandatory level.
The ultimate aim of the mandatory private pension system is to provide anuities (pensions) starting from the standard retirement age (65 for men, 60 for women). Till then this system only accumulates money from contributions. The costs of an old age annuity depends on:
a) yields obtained from investing the lump-sum accumulated up to retirement;
b) life expectancy of a person who is entitled to this annuity. Life expectancy is related to the person's age and sex.
The enhanced role of the private sector in pension provision has led to substantial research on the performance of private systems, but most of the analytical effort has been focused on the accumulation phase of private pension provision. In Romania there are very few analyses of the payout phase, which involves the transformation of the final balance into retirement income through annuities. The main concern in those analyses is the interest rates (yields): the risks related to investment rates, the influence of the interest rates upon the accumulated sums up to retirement and the direct relationship between interest rates and annuities.
Recent years were defined by low rates and the expectations are that interest rates in the future will rise. When we are in a low interest environment, future benefit payments are worth more in the present then they are when interest rates are higher and vice versa. Therefore, any factor that affects the duration of future payments is more than significant.
In a study published in the Journal of Epidemiology and Community Health in 2002, professors C. Dolea, E. Nolte and M. McKee present many improvements in medical care in Romania, and conclude that Romania is at last beginning to follow the path of improving adult mortality rates (increasing life expectancy) seen in the early 1990s in some of its neighbors.
By analyzing the past, actuaries today can predict future real life outcomes with surprising accuracy. The process of evaluating past data, assessing the factors involved, and modeling the future is called "actuary mathematics". By this method, professionals are able to determine life expectancy at different ages based on past results, and even predict the probability of their death in the following years. This complicated statistical and mathematical system requires accurate data, to obtain death probabilities and predict the most likely outcome for the future.
A natural consequence of increasing life expectancy is a longer period of retirement and therefore a need for a longer period to pay pension benefits. Individuals are becoming more concerned about their ability to afford to live out those extra years in comfort, and longer pension commitments are placing increased burdens on limited resources.
A growing concern worldwide is that improvements in life expectancy are not being adequately allowed for in pension calculation and therefore accounting valuations of the pension funds.
While it is impossible to predict when any particular individual will die, if we have a very large group of people and know certain basic information about them, such as where they live and their age and sex, we can predict the number of survivors each year. However, life expectancy is not static: past and often future, improvements - in medicine, standards of living etc. - need to be reflected in these predictions so that they continue to be reasonable.
Predicting survival rates far into the future will always be subject to error, as changes in improvement trends are likely unknown. To highlight the time horizon for the survival assumption, a 20 year old employee earning benefits today may still be receiving pension payments when he or she is 90 years old, and his survivor many more years after his death (if he has chosen a combined pension: old age and survivor).
Another consideration is that survival experience could be affected by unpredictable local or national events, such as storm disasters, avian flu pandemic, which would not have been included in the tables of assumed or observed survival rates.
2) The changes in life expectancy in Romania and other developed countries in the last five decades.
In table 1 we present the variation in observed life expectancy Romania and compare it with some developed countries. In all countries, due to higher standards of living - which includes, among other things, better medical care and better nutrition - mortality rates are in a declining mode. Life expectancy at each age is increasing at the rate of more than one year in every ten years. With the average period of retirement in those countries now a little under 20 years, one extra year of payments can add up to 5% (one-twentieth) to the cost of benefits. Considering that some pension funds will use survival rates tables that may underestimate future life expectancy by several years, these additional costs can soon add up. To date the Romanian supervision authorities have not regulated this field.
The variation between developed countries can be significant. Residents of United States have a life expectancy some five years shorter than those in Japan. This variation makes it important to consider the geography of the population to which the assumptions are to relate. Developed countries publish on a regular basis new tables of survival rates. These tables are based on pension plan mortality experience collected over a period of several years.
Separate tables are available for males and females, white and blue collar employees, and also for able-bodied and disabled participants. While the distinction in mortality rates between males and females and between able-bodied and disabled participants is apparent, the distinction between white and blue collar workers is less obvious. This is an example of the "income effect", whereby larger mortality improvements have been observed among wealthier individuals. Therefore, a scheme covering blue collar workers, white collar staff and executives may consider using different survival assumptions for each group. Alternatively, income thresholds (for salary and pension) can be used to distinguish between different standards of living, with higher survival rates associated with higher incomes.
Table 2 (males and females) presents the observed mortality rates in Romania in the years 1991-2003 for the entire population (rural and urban, active and disabled people).
These data were prepared by the RNIS and are the only reliable data existing for the Romanian population for those years.
It can be observed that the same phenomen of decreasing mortality rates occurs in Romania with a higher intensity as of the year 1990.
The mortality rate in the urban environment is different from that in the rural environment. In the rural environment the mortality rate is higher: life expectancy at birth for men in the year 2003 was 66.4 years in the rural environment and 68.2 years in the urban environment. For women it was 74.1 years in the rural environment and 75.4 years in the urban environment.
The decline in the mortality rate is not equal for each age. For ages under 40 the decrease in the mortality rate is greater than for ages higher than 40 as a result of higher standards of living during their lifetime (better food, better medical care, etc).
3) Adapting the existing Romanian mortality tables to future changes in life expectancy.
Projection for future improvements in survival rates can be made either for all ages equally or, arguably more appropriately, varying the level of improvement for different generations. Survival tables that do not account for generational differences are called "static" tables. Static tables do not account for the different survival rates that a 20 year old may see in future retirement compared with a current retiree, although experience highlights significant differences between these generations.
Sometimes life insurance companies incorporate general improvements in mortality by way of an "age setback". An age setback is a way of treating people as if they were younger than they really are. For example, using a five year setback for a 60 year old would imply the use of survival rates at age 55, therefore assuming a longer expectancy.
For pension providers we propose to take into account future changes in the Romanian population mortality rates, using a combination of age and year of birth on which to base an individual's life expectancy. This pattern also introduces allowance for what is referred to as the "cohort effect". The youngest ages will show the largest improvement in life expectancy, as projected improvements will be greater for this group under the new tables.
From the year 2003 on, the presumption is that the mortality rate in Romania for the active population will improve exponentially by 0.5% per year. This improvement is expressed by Equation no.1.
Based on this equation we prepared for each birth year a table in function of age and gender of mortality rates till age 100.
So, in table 3 we present the mortality rates and life expectancy in year 2008 for ages 18 to 100. We use this table to know the number of alive persons and for estimating the costs of the annuities the members of the pension funds will be able to buy at standard retirement age. More persons from the future generations will survive and receive an old age pension than from the generations which have already joined the new private pension system. At the same time, the existing and future generations will expose the pension providers to the longevity risk due to declining mortality rates and rising life expectancy.
The mortality rates measured by the Romanian National Institute of Statistics is at each age for the entire population at that age. There is no differentiation between active people (able to participate in the work force) or disabled people (integral or partial).
New professional studies about the mortality rates of the different segments of the Romanian population (urban, rural, active, disabled) are needed. We consider as a proper evaluation that the mortality rate for the active and working population in 2003 was 80% of the mortality rate measured by RNIS for the entire population (flat rate for each age). For the retired population - age 60 years on - 75% of the mortality rate measured by RNIS (flat rate for each age). Those percentages are assumed after analyzing the percentages included in the EVK2000 mortality tables (tables used by the Swiss insurance and reinsurance companies).
Invalidity is defined as a permanent or partial incapacity to perform a paid activity due to an illness or accident. The mortality rate of the disabled population is greater than the mortality rate of the active population .The probability of surviving till the standard age of retirement (men 65 women 60) is different for active persons and for disabled persons. Due to the difference in the mortality rates (life expectancy), the cost of an annuity (life time pension) at the standard age of retirement differs for active persons and for disabled persons.
4) Estimates of the costs of future improvements in Romanian mortality rates
The formula for a single immediate life annuity issued to a person aged x is as set out in equation no. 2.
Providers' costs include operating costs and commissions paid to annuity brokers. Those costs are influenced by the existence (or non-existence) of a competitive annuities market. In this paper we avoid those costs because they do not affect the thesis and the conclusions.
A single premium payment based on a risk-free interest rate (measured by the interest rates on government or central banks securities) is interpreted as an actuarially fair annuity. Our calculations assume investment yields of 5% which include a certain degree of risk.
Table 4 presents the cost of a single premium payment for an annuity contract which provides a monthly payment of 1000 Lei. The real interest rate along "time" is fixed, and only the mortality rates for each age are changed as follows:
a) First, by multiplying the Romanian mortality table as calculated by RNIS for the entire population - starting with the standard retirement age - for the year 2003 by 0.75. Any other coefficient that is chosen (including 1.0, which means that there will be only the adjustment mentioned in point b) below) will not affect the conclusions concerning the influence of the increase in life expectancy presented in section 5.
b) Second, by adjusting the mortality rates according to equation 1 (for the year 2008 - t=5) for each age.
The annuity costs based on RNIS mortality tables for year 2003 are for active population so the mortality rates for each age were calcullated as it is mentioned at point a) above ( without taking into consideration future improvements in mortality rates).The annuity costs based mortality tables calculated according to the model proposed (for year 2008) include also the calculations mentioned in point b) above.
5) Conclusions from the data presented.
For men at age 65 (in the year 2008) the cost of an annuity based on mortality tables calculated for the year 2008) is greater by 2.5% than the costs based on the mortality tables published by RNIS for the year 2003. For women at age 60 the difference is 2.1%. Life expectancy for men at age 65 is 77.6 years For women at age 60 life expectancy is 78.3 years. There remains a very short period in which the improvement in mortality rates can have an effect.
For men at age 30 (in the year 2008) the costs of an annuity payment of which commences from age 65 based on mortality tables calculated for the year 2008 are by 9.3% greater than the cost based on the mortality tables published by RNIS for the year 2003. For women at age 30 (in the year 2008) the difference is 5.7%. Life expectancy from age 65 (for men at age 30 in the year 2008) is 79.5, 2 years more than the life expectancy of men at age 65 now, in the year 2008. For women at age 30 (in the year 2008) life expectancy at age 60 will be 80.2 years - also 2 years more than the life expectancy of women at age 65 now, in the year 2008.
For young people the period in which the improvement in mortality rates has an influence is longer, besides the fact that due to the improvement in mortality rates more and more people belonging to those generations will reach the standard retirement age 65 for men and 60 for women.
As shown above, the increase in annuities costs due to the increase in life expectancy are very substantial: between 2.1% for men and 2.5% for women who are already at standard retirement age, 5.7% for men and 9.3% for women at age 30 and 11.3% for men and 6.9% for women at age 20 .
Questions which arise are: a) how do we hedge mortality risks; and b) who will bear the costs of paying pensions for a longer time.
a) Hedging mortality risk.
Having now discussed the survival assumption in detail, we can turn our attention to the risk of future survival experience not being in line with our assumptions. It is already possible to use financial instruments to "hedge" or reduce investment risks associated with future events. For example, a pension fund may already use options or swaps to immunize itself against unwanted investment risks.
This is not the same as insurance, as the plan maintains control of tolerable risks, hedging only those to which it does not wish to expose itself. Would it be possible for a pension provider (pension fund, insurance company) to mitigate the risk associated with changes in future life expectancy, without the large cost of securing the benefits directly with a reinsurer? The answer - in theory - is affirmative.
Mortality-linked securities and over-the-counter swaps both aim to provide a solution, although neither method is straightforward. One such available security is a longevity bond. Although there are several varieties, the "classic" longevity bond pays a regular coupon (like an annuity) based on the number of survivors at each future age from an original covered population. Payments continue until the last survivor dies. As these products are further developed, we are likely to see them being used by pension funds keen to reduce their exposure to future changes in life expectancy.
In practice there is an attempt to resolve the financial problems caused by longevity by raising the retirement age. We can expect that if the increase in life expectancy will seriously distort the financial sources of the public system and/or significantly reduce the pensions that the mandatory pension funds will be able to provide, the standard retirement age will be raised. Changes in contributions are not usually effective. Contributions cannot be collected for the period already passed, but only forward.
b) Bearing the costs of paying pensions for a longer time.
Who bears the cost of these longer pensions depends on who has the obligation to make the future benefit payments. For traditional public systems based on defined benefits plans, it is the State (indirect tax payers) that is exposed to the risk of benefits being paid longer than expected.
Public pension systems balance between an increase in pension payments without a parallel increase in income from contributions to social security system by:
1) changing level of contributions
2) erodating pensions by inflation
3) increasing the retirement age or
4) a combination between the three.
The same way they are expected to deal with the additional payment pension costs due to the increase in the life expectancy.
While increased disclosure and reporting requirements for traditional pension arrangements have led to an environment where even well-established plans are moving away from defined benefits, added costs from longer pension payments have also contributed to reviews of plan design.
In Romania the mandatory and facultative pension funds operate defined contribution plans, whereby company and employee contributions are paid into an account which accumulates with investment returns until retirement. The obligation to the employer is only for the specified contributions, not the ultimate benefits. The risk of unknown future payments - for those contributions - has been transferred from the state to the employee. This is where the risk remains, with the employee being responsible for budgeting the fund over his or her remaining lifetime. If the individual uses the fund to buy an annuity from an insurance company (in Romania for the mandatory and facultative pension funds the legislation is not yet completed and we do not know if it will be allowed to transfer the accumulated sums and buy annuities from insurance companies), the risk of living longer than expected is then transferred (at a cost) to the insurer.
The assumptions for life expectancy used by insurance companies will therefore have a fundamental impact on the level of pension that can be afforded in retirement.
1) Actuarial Mathematics (Second Edition), 1997, by Bowers, N.L., Gerber, H.U., Hickman, J.C., Jones, D.A. and Nesbitt, C.J.;
2) Dolea, C12; Nolte, E2; McKee, M 2- Research Reports, Journal of Epidemiology & Community Health,56(6):444-449, June 2002;
3) Michael M. Parmenter, Theory of Interest and Life Contingencies With Pension Applications-A Problem Solving Approach, Actex Publications, March 1999, ISBN 978-1566983334;
4) Policy Research Working Paper 4438, Annuities in Switzerland, Monika Butler, Martin Ruesch, The World Bank Financial Systems Department, Financial Policy Division, December 2007;
5) Robert W. Batten, FSA, Life Contingencies: A Logical Approach to Actuarial Mathematics, 2005 Edition, ISBN 1-56698-522-6;
The most recent stories on this subject:
Query failed: SELECT ac.ID_comentariu, ac.alias_utilizator, comentariu, DATE_FORMAT(ac.data, '%d.%m.%Y %H:%i') AS data_comentariu, ac.vizibil, u.nume, u.prenume FROM site_articole_comentarii ac
INNER JOIN 1asig.site_utilizatori u ON ac.ID_user = u.ID_user
WHERE ac.ID_articol = '220'
AND vizibil = 1
ORDER BY ac.data DESC