Early Retirement in the Government Sector in Egypt: Preferences, Determinants, and Policy Implications
This study looks at factors affecting the decision of early retirement for Egyptian government sector employees. The empirical analysis is based on 2005 nationally representative sample of government sector workers. Among the findings of this study are: women are more likely to retire earlier than men; good health status is associated with longer stay on the job; the level of education is not a determining factor for women, but it is for men; men plan to work after their early retirement; the presence of the working wife has a positive effect on her husband to retire early, yet a working husband discourages his wife from retiring early. Policy implications discussed timing of the announcement of the plans; the potential outcome of excessive payouts, as well as the need for the government to provide early retirees with necessary skills and training to plan their after retirement years.
J18, J26, J33, C20, H55
Retirement, Employment, Government Sector, Egypt
One of the axioms of stabilization and structural adjustment policies suggested by the World Bank and the IMF is privatization and downsizing of the overcrowded public and government sectors. Early retirement has been considered one of the most effective tools of downsizing. Often financed or heavily subsidized by the government, early retirement facilitates the privatization process, and mitigates the adverse social impacts of layoffs when unemployment benefits are unavailable. In 1990, Egypt embarked on a program of economic liberalization and reform. At the heart of this program has been privatization of state-owned enterprises (SOEs), public and government sectors. The rationale for privatization and labor force restructuring is to fuel private sector interest in the businesses being privatized, such as SOE. Shedding excess labor would render the business more attractive to prospective investors, and reduces operating costs to enhance the business’s financial viability. [End Page 79] Privatization and structural adjustment proponents believe this process would not be efficiently realized without an effective, voluntary early retirement program, with a satisfactory and fair compensation packages1. An arrangement that fairly compensates workers for the loss of guaranteed lifetime employment and benefits. This instrument— early retirement incentives-- is often used where labor laws ban layoffs, or where labor unions are strong. Based on the proceeds generated from privatization in Egypt, a fund was created in 1997 to implement early retirement programs.
One of the features of the Egyptian economy is the dominance of state-owned enterprises, which has been characterized by overstaffing and excess employment. According to CARANA (2002a), estimates of overstaffing in typical SOEs before labor force restructuring range from 30% to 60%. At the start of the privatization program in 1991, Egypt’s total labor force was around 15,250,000 employees. The government was by far the largest employer in the country, accounting for 37.4% of aggregate employment, around 5,500,000 employees (CARANA (2002b)).
The minister of social insurance, the minister of finance, and the minister of investments in Egypt have put pension reforms high on their reform agenda. Before the new reform package, which went into effect on 15 January 2006, the retirement age for women was set at 50, the same as for men. Under a new reform package, women are eligible for early retirement at the age of 45 provided they have worked at the same enterprise for at least 19 years; whereas men are still eligible at the age of 50. Women’s groups believe these reforms permitting optional early retirement for women at an earlier age than men will result in increased marginalization of women workers. On the other hand, Egyptian economists say it will grant companies greater freedom and promote privatization. Al-Ahram (a daily newspaper) praised the reform as being “useful,” reporting that the government had earmarked US $80 million for the scheme.
Although several studies have discussed retirement benefits and related problems in Egypt such as the the need to reform the PAYG (Pay As You Go) system in Egypt; and the different investment strategies regarding the fund assets (Maait and Khorasanee (2000) and Osman and Salah, (2001), this is the first study dealing with early retirement preferences for government sector workers and the determining factors in the decision to retire early. [End Page 80]
The purpose of the study is to identify variables describing the decision to retire before the age of 60, the legal age of retirement, and characteristics of those early retirees. With many employees accepting early retirement packages in Egypt, the need to understand the factors determining their decision is relevant.
This study aims to answer the following questions:
Is early retirement a preferred choice for government sector workers?
What are the characteristics of workers expected to retire before the legal age of retirement?
At what age bracket is the worker most likely to retire?
Is wage level a determining factor in early retirement decision?
Are double wage earners in the household more likely to retire early?
Is the level of education a determining factor in the decision to retire early?
The rest of the paper is organized as follows: Section II is a review of the institutional background and early retirement reforms in Egypt. Section III offers a literature review on the methodology. Section IV presents data, methodology and estimation procedures. Section V discusses empirical findings of the study. Finally, Section VI provides conclusion and policy recommendations.
Institutional Background and Early Retirement in Egypt
The current social security system in Egypt was established by the Law 79 of 1975 which covers civil servants and employees in public and private sector enterprises. The system was then extended to include the self-employed (Law 108 of 1976), Egyptian workers abroad (Law 50 of 1978) and finally embraced casual workers (Law 112 of 1980). Law 79 of 1975 provides benefits for old-age, disability, death, work injury, illness, medical care, maternity and unemployment. The social security system is managed by two separate Funds: one manages Government workers and the second handles workers in the public and private enterprises, the self-employed, casual workers and Egyptians working abroad.
Early retirement is optional for employees, who wish to end their service before the legal retirement age of 60 years. It is administered with the cooperation of the holding companies, their affiliates, the Egyptian labor syndicate union, and the employee. The employee must have at least 20 years of contributions. If the insured is under age 45, the pension is reduced by 15% and by 10% if the worker is between 45 and 49 years, and by 5% if between the ages of 50 and 54. The minimum pension is 50% of the average wage in the last 2 years (if the qualifying period is not less than 20 years) or 100 pounds a month. The maximum pension is 80% of average earnings or 920 pounds a month, whichever is less. Lump-sum gratuity: 15% of the average annual basic wage for each year of contributions beyond 36. (SSPTW (2003)). The size of the pension in the early retirement incentive plans depends on the worker’s age, the number of working years, and the average salary over the last years.
Early retirement age for both men and women was set at 50. Under the new reform package which went into effect on the 15th of January 2006, women will have the option to retire at age 452 if they have worked for the same organization for at least 19 years, and receive a maximum compensation package of $7,500 and a minimum of $2,900. Critics and women’s groups believe that current reforms permitting optional [End Page 81] early retirement for women at an age as early as 45 will result in increased discrimination for women workers. They suggest that women employees who have an earlier access to lump sum money before men, and women in low income families will feel pressured to volunteer an early retirement to pay off family debt or help in household major expenditures (VOA (2006)).
Review of the Literature
Feldman (1994) adopted a traditional definition of retirement, and was followed later by others such as Beehr et al. (2000). According to Feldman, retirement is defined as a state of exit from a current job that has been held for some time, with the intent of less commitment to work and a decision that is taken sometime after the individual’s middle age. Dutcher (1999) also has provided several definitions for retirement: an individual may exit his career out of the labor force; may begin a new career; work part time in his previous organization or work part time in a new one. It may also encompass other characteristics such as: it is a sudden, rather than a gradual, process; it is a permanent state; and in case of early retirement, it is a voluntary choice, even though made subject to opportunities and/or constraints offered by employers and pension arrangements. Previous research on early retirement in developed countries in general and on women in particular is limited, for lack of data. There are three common elements in the recent trends of labor markets across developed countries. First, all countries face aging of their population and labor force. Second, in all developed countries, the labor force participation of older workers has decreased significantly in the last few decades. Third, most governments of developed countries stress the importance of reversing the trend toward earlier retirement from the labor force, as it is the case of European Union (de Vroom and Guillemard (2002), and Herbertsson (2001)).
Nearly all developed countries have a standard retirement age of 65 years. Although steps are taken to raise the retirement age to 65 gradually, Japan, Germany and women in the UK still have an official retirement age of 60 years. The US has decided to increase the age of entitlement to full benefits beyond the age of 65. Sweden has proposed a law to give the employees the right to remain employed until the age of 67. Moreover, in Sweden and the US, mandatory retirement arrangements are abolished. Guillemard (1999) states that the above measures signal a determination of governments to put an end to the golden age of early exit and to gradually shut down retirement schemes.
The literature on the effect of the economic incentives on retirement treats the retirement decision essentially as a labor supply issue. Workers who approach the retirement age evaluate their prospective wage and pension streams, and choose the retirement age that maximizes their expected lifetime earnings or utility. Substantial empirical evidence indicates that the incentives provided by the social security systems have an impact on the age of the labor force withdrawal. Pensions that are actuarially unfair encourage early retirement, and countries with more generous social security benefits tend to have a lower average retirement age (Gruber and Wise (1998)). Boskin (1977) was one of the first to pay close attention to the effects of incentives on early retirement. Others followed suit were: Quinn, Burkhauser, and Myers (1990); Stock and Wise (1990) and Fields and Mitchell (1984) in the US. Empirical work in Europe has also [End Page 82] examined early retirement using an incentive-based approach; examples include Borsch-Supan (1992) for Germany and Meghir and Whitehouse (1996) for the UK. Employer (demand) side on the other hand has received much less attention in the retirement literature. In a pure labor supply model, workers are free to choose the retirement date that is optimal for them. Yet firms may also encourage their workers to retire early. Early retirement can be a soft way to reduce or to renew the workforce. Godofsky (1988) found out that downsizing organizations in the US induced early retirement by offering incentives such as five extra weeks of severance pay and five extra years of pension in addition to calculating the pension as if the individual is five years older.
The bulk of the research on early retirement has focused on individual differences variables (such as gender, marital status, health status, work related factors such as years of experience and level of education, and attitude towards retirement), and their impact on the decision to retire (Beehr (1986); Pollman and Johnson (1979); and Schmitt and McCune (1981)). Regardless of the important gender differences, most of the research on early retirement is studies of men (e.g. Haveman et al (1988); Berkovec and Stern (1991); Blau (1994); Meghir and Whitehouse (1997); and Riphahn (1997)), and the limited available research on gender differences do not lend a clear evidence as to who is leading who.
One class of research suggests that men as a group retire earlier while women as a group do not (Clark (1988); and George et al (1984). Ruhm (1996) suggested that since women normally marry older men, management of retirement implies that wives are likely to retire at a younger age than their husbands. Besides, Erdner and Guy (1990) confirmed that individuals were less likely to retire if their spouses were working. The presence of children in the family appears to have only small effects for women, while it significantly reduces the probability of leaving the labor force for men (Perrachi and Welch (1994); see also Reitzes et al (1998)). In their study on gender differences in retirement decisions in a large manufacturing organization in the US, Talaga and Beehr (1995) show retirement decisions differed between men and women primarily when dependants lived in the household, when one’s spouse was retired, and when the health of one’s spouse was a consideration. Poor health in general contributes to early retirement decision (Muller and Boaz (1988)) and makes it a necessity rather than a voluntary decision. Not all health conditions will influence the decision in the same way or direction. Anderson and Burkhause (1985); and Colsher et al (1988) found that poor health conditions will positively influence the retirement decision, but a moderate health condition may influence workers to stay longer in the labor force.
Research on the effect of education on early retirement decision showed an ambiguous relationship as well. On the one hand, higher education increases the opportunity cost of leaving work, given that education is closely related to earned income. On the other hand, accumulated savings from higher wages throughout the worker’s life may double as another proxy for wealth and act as an incentive to retire early. Besides, as Quinn et al. (1990) suggest, for the United States, education may increase non-monetary benefits associated with work. These effects will act to prolong the working life of the more educated worker. Other studies addressed an important set of variables that influence early retirement decision and related to the opportunity for different career tracks. Variables such as secondary vs. primary jobs, and part vs. full time jobs. (Doeringer (1990); and Ruhm (1990)). A third set is related to organizational [End Page 83] variables such as wage level, pensions, expected benefits, type of industry, sector and size of the firm (Rosen and Jerdee (1989)).
Results emerging from the Nordic countries were not conclusive either when it comes to gender differences in retirement decisions. Using data from Denmark and applying a competing-risk model with three different end-states, Pedersen and Smith (1996) find significant gender differences in the decision to retire early. These results may be contrasted with those of Lilja (1996), based on Finnish data and using a competing-risk model with four destination states, who finds the propensity for early exits does not differ significantly between men and women. The study also shows that the presence of a retired spouse at home encourages the other spouse to consider early retirement. Of the limited literature on the issue of early retirement in developing countries are those of dismissed workers from SOE in Turkey. Tansel (1998 SOE in Turkey. Tansel (2003) estimates of the impact of privatization on dismissed workers in Turkey in the cement and petrochemicals sectors, and evaluates welfare losses associated with moving from high rent formal public sector jobs, to informal and poor quality private sector jobs. About one third of the dismissed workers’ sample continued to work, where half of them established their own business with the help of the severance package.
A recent survey in Egypt was conducted to determine the consequences of accepting early retirement after privatizing several public sector enterprises, showed that many retirees quickly spent their compensation on family emergencies and other immediate needs, the reduced pensions were not adequate to support their families, and they did not obtain new jobs. Retirees became unproductive, and burdened their families. The survey pointed out to the shortfall of the program to meet social objectives for early retirees. Workers did not have sufficient information to make a rational decision, nor were provided by training or counseling services. In many instances, workers reportedly were pressured to volunteer for early retirement suggesting that remarks were made to them that: they may receive poor treatment from new owners; may face relocating to inconvenient site; government resources were limited and so any deferral of a decision could result in lesser compensations in the future; …etc. On another front, workers eligible for a pension did not understand that early retirement pensions are reduced pensions. Others did not expect changes in the availability of health benefits. (CARANA, (2002a)). Akeel (2000) stressed the hardships women face due to early retirement, such as deprivation of free medical treatment, the value of the meal provided by the public enterprise, and deprivation of their share in profits and incentives.
Data and Descriptive Analysis
The empirical analysis is based on 2005 nationally representative sample of government sector workers. The survey was conducted by the IDSC (Information and Decision Support Center in December 2005). The sample resulted in 3437 current employees of government sector personnel, ranging in age between 50 and 57, covering seven governorates: Cairo, Giza, Behira, Sharqia; Suez, Asyout, and EL-Menya. The sample is representative of all three professional ranks and activities (administrative (Idary), (Hayaat) and local (Mahaleyat))3. The final working sample resulted in 33974 government employees. The survey provides rich data on basic demographics, experience, occupation, hours of work, wages, health, income, current levels of expenditures, skills, [End Page 84] attitude towards early retirement; and planned investment after retirement. In what follows, the word “expected” is synonym to the word “preferred”. That is, expected age of retirement refers to the age at which the respondent expects (or plans) to retire, not the predicted age of retirement.
In the following analysis, the preferred age of retirement is arranged into three age brackets: expected to retire on or after 50 but before 55 (i.e. 50–54); expected to retire on or after 55 but before 58 (i.e. 55–57); and those who don’t plan to retire early are grouped into “old age” category (i.e. retiring at 60 years of age, the legal age of retirement). Figures (1 and 2) below, and Table (A-1), in the appendix, present the distribution of respondents by their expected age of retirement and gender. Among the most important statistics are5:
• About 70% of the sample is expected to retire between 50 and 57—see figure (1), with almost equal percentage for women and men (36% vs. 34% respectively).
• Two thirds of women expressed their desire to retire between 55 and 57 are currently the youngest of age (i.e. between the ages 50–53, compared with 56% for men). Men exceed women in their preference to retire later. Over a third of men plan to retire at age 60, compared with 22% of women (figure 2).
• Most of the respondents, especially women, have over 20 years of experience (i.e. they are eligible for early retirement if they satisfy their contribution conditions).
• Nearly all of them are working in urban regions.
• More women are well educated than males, with at least university education, and close to a third, and in some cases over a third, of men are found with primary education or lower6.
• A little over one fourth (26%) of women preferring to retire at old age are the primary breadwinner of the family, which may explain their preference for late retirement.
• Nearly all men in the three categories are the heads of the household, while a sizeable portion of women preferring to retire early, are spouses to heads of their household.
• Considering the professional level of the current sample, the table reveals that over two thirds of women in all retirement categories hold the highest occupational ranks7(first and beyond), which may explain the higher wages they receive over men.
• On average, men earn 82% of women’s earnings, which complements the fact that males, on average, hold lower positions/occupational ranks than females, as indicated in the previous point.
• The majority of workers has between three and five kids in their household, working women however, have lower number of kids (3–5) than males (over 5 kids).
• House payments do not constitute a major expense since the common trend is either own the household or pay an old rent.
• Between 59–65% of men in the three retirement age categories have an out of the labor force spouse, in contrast, between 65%–85% of women have working spouses. [End Page 85]
[End Page 86]
• One surprising outcome though is when asked if spouse would work after retirement, 84% of women retiring early (i.e. at 50–54) are not expecting their spouses to work after their early retirement. Then again, close to two thirds of men preferring to retire as early as between 50 and 54 of age, are expecting their spouses to work, at least part time. Interesting enough to find out that most workers believe their current work does not require any skill; and over 75% of them report that they don’t use computers at work.
• It is also obvious that women are eager on receiving new skills when over two thirds of them reported they received training on new skills, relative to less than 50% of men across the board.
• When asked about sources of income after retirement, an average of 88% of all respondents reported their pension income is their main source; over half of that money will go to help support their kids’ education or their marriage expenses. An average of 12% is saved for Hajj or tourism, whereas an average 13% was reported as “not decided yet”.
• Across the board, over a third reported they expect their spending to increase after retirement, probably reflecting their awareness of increasing level of prices in order to stay at the same level of living.
• On average women and men are equally likely to view their health as either good or excellent and a higher proportion report the same when it comes to the health status of their household members, as well as benefiting from their health insurance.
• When asked about their expectation to work after retirement, the descriptive statistics indicate the majority of women expect not to work, whereas men’s preference to work after retirement dropped with increased age of retirement. For example, 61% of men expecting to retire early (50–54) plan to work after retirement, 53% of men retiring between 55 and 57 plan to work after retirement, relative to 45% of those who plan to retire at the age of 60, inevitably fighting over limited jobs with other unemployed and predictably raising the unemployment rate.
• Over 50% of men planning to work after retirement expressed their interest to take on a private venture. Except for women preferring an old age retirement, a large percentage of workers (men and women) plan to work in their own private venture. Additionally, the majority of them don’t believe they need any training to work after their retirement. This is an important statement to pause at and analyze further. Of all women in the sample, only 23% reported that they need a special skill in order to start a new work after retirement, compared with only 13% for men. Simply put, most of the workers don’t believe they need special training or skills to run their own business, or start a new job. Lack of skills necessary to run these businesses normally result in partial or total failure, adding further to the unemployment lines.
• When asked if they already know about early retirement plans, a sizeable number reported they don’t know anything about the policy (40%–60%), whereas the media was the source of knowledge for almost a third of who reported some knowledge. In most cases, less than 10% of workers know about early retirement plans through their jobs. [End Page 87]
• Finally, the most important reason of not accepting early retirement has been the declined income after retirement.
In summary, men are less educated than women, therefore, occupy lower ranks and consequently earn lower wages. Men are also the heads of their households, and are the single wage earner, whereas women are the secondary wage earners in their households. Unlike women who received training on new skills, men did not acquire any skills, but planning to work after their early retirement. In the end, it also seems that only women planning to retire between the ages of 50 and 54 are familiar about the new early retirement policy.
Methodology and Empirical Findings
In order to assess the effect of personal and job characteristics on the decision to retire, controlling for other covariates, the probability of planning to retire at a certain age bracket is estimated using a multinomial logit model where the decision to retire is one of three choices: retire between the ages of 50 and 54; retire between the ages of 55 and 57; or retire at the legal retirement age (60 years).
The functional form of the multinomial logit model (Maddala, 1983) is employed as follows:
= 0 for expected old age retirees
= 1 for expected retirees between the ages of 50–54
= 2 for expected retirees between the ages of 55–57
The probability that the ith individual selects the jth working status is:
Where the subscript j=0, 1, and 2 is for retirement age, Xi is a vector of independent variables (region of residence, age, years of experience, level of education, relation to head of household, worker’s health status, and health status of the household; professional rank, expectation to work after retirement, work status of spouse, current wage, students present in the household, number of household members, presence of working members in the household, additional income in the household, and prior knowledge of early retirement policy), and βj is the parameter vector for retirement age category (j).
To obtain the marginal effects of a covariate, xi, on the choice probability to state j, Pj, is given by:
Where βj and βk are the relevant elements of the parameter vector β. (Greene, 2003). [End Page 88] Thus, the marginal effects are to be interpreted as the change in the probability of ending in a particular state j given a change in an explanatory variable xit.
The appropriate theoretical framework for the analysis of choice of retirement age and its determinants is the life cycle model. Such a model would specify how decisions about education, labor supply, income, health and consumption are made.
Generally, life cycle models assume that people save in order to meet up their financial needs during retirement. People may borrow when young, save when middle-aged, and spend their savings when retired to level off their consumption over the life cycle. In a simple life cycle model, consumers are assumed to view the future clearly. Therefore, in making their plans, workers are assumed to know how much they will earn over their lives, when they will retire, and their expectation for consumption shocks.
The decision to retire early is one of a set of joined decisions made over the individual’s adult life. Furthermore, it is not necessarily a free choice. Early retirement (or exit from the paid labor market) can be forced upon the older worker by, for example, unpleasant labor market conditions; own health problems or of a close family member. Though ignoring other factors gives the study a narrower perspective, available data is an immediate constraint.
Variables Used in the Model
People do not make retirement decisions based solely on economic factors. Social, personal, as well as economic variables may be important in the acceptance decision. Therefore, this study includes vectors of these variables.
A multitude of both individual levels “push factors”--attitudes towards retirement, work history and health status-- and “pull factors”—added worker effect, alternative employment, expected pension wealth-- all enter into the worker’s decision about early retirement.
Age is included in the analysis. It is expected that early retirement decision decrease as the individual ages. The number of years of experience in the current institution may be of more importance at predicting early retirement decisions. Workers who spent long time in one place are more likely to have accumulated higher wages and therefore accrue higher pension benefits. In contrast, those who have less career service are likely to have less pension benefits (Kilty and Behling (1985)). Therefore, they are less likely to voluntarily retire early. For women, especially, marital status is important since unmarried, widowed, or divorced women tend to be financially disadvantaged. A dummy variable for being married is included. [End Page 89]
The level of education as well, influences the decision to retire at a certain age. As it is the case in government sector employment, income and non-pecuniary benefits are associated with the level of education. Therefore, higher education may increase the opportunity cost of leaving work. Investment in higher level of education implies that the individual intends to work more over his lifetime and retire later than others who chose not to make such an investment. Health also is a determining factor. If the individual anticipates declining health with age, he may compensate for that by saving/working more and retiring early.
Wealth depends on earnings ability, in addition to income of other members in the same household. The influence of wealth on the retirement decision is not clear. For example, increased wealth is expected to improve the possibility of early retirement through increased ability of independence. Nonetheless, wealth may be a proxy for both ability and social status, in which case, it is expected to reduce the probability of early retirement. Number of other working members in the household is added. Another dummy representing the presence of students in the household is also included.
To test for the added worker effect, spouse’s employment status is added. It is possible that the decision to retire for working couples is dependent on their spouses’s working decision. However, married women may well feel more pressured to work as they approach the legal age of retirement if their spouse is finding it difficult to work, and less pressure to work if their spouse is comfortably off. In general, a higher level of income, induces higher level of consumption and allows the retirement at a younger age, however, there are some complications in this simplistic view, such as uncertainty of [End Page 90] receiving a constant income in the future. Likewise, two income couples are more likely to have higher accumulated savings and pension benefits. Rank is likely to show the individual’s income earning capacity over his working life, rather than at the point in time of the survey. Therefore, it may also be used as a proxy for wealth. The higher the rank, the more likely the individual has been accumulating wealth over his work life. Finally, prior knowledge of the early retirement policy may influence worker’s decision for early retirement.
Upon close review, however, it becomes evident that many of the previously mentioned assumptions can be at odds, and the worker could fall into either category. For example, it is hypothesized that workers with lengthy years of experience may prefer to retire early since their pension benefits and savings could be high enough to comfortably make such a decision. On the other hand, long years of experience may signal close ties with the work place, and the worker is less likely to retire early. Another example is dual earning families, who may have accumulated sufficient savings for one of them to accept early retirement; at the same time, children expenses may become the reason for deciding not to. Although wages in the public and government sectors are gender neutral, women represent a large percentage of this sector, and they normally retire at a younger age than men. Therefore, it is important to distinguish between men and women in the decision to retire at a certain age. In this analysis, I allow the covariates to have various impacts on the flow to different states for the two genders by carrying out the analysis separately for men and women. Using “stepwise” regression, the model is rationalized into the most significant variables affecting the decision of early retirement.
An understanding of the factors that determine the decision for early retirement provides an insight for policy makers in their planning decisions. In this section, results of the marginal effects of the multinomial logit model are presented. Marginal effects represent the change in the probability of ending in a particular state given a change in an explanatory variable. The set of explanatory variables depends ultimately on the available data and the reliability of the data. The analysis focuses on three groups, henceforth denoted as retirement at: 50–54; 55–47 or Old Age (60) 60 years.
A reference man in this model is 53 years old; resides in an urban neighborhood; with 25 years of experience; not married, with an average monthly wage of LE 701; a secondary or less graduate; with poor health; does not benefit from health insurance; occupies a low professional rank (fourth or third); has a non working spouse and two [End Page 91] students in his household; his monthly expenditures is 62% that of a comparable woman with the same characteristics; doesn’t plan to work after retirement; and doesn’t know about early retirement plans. A reference woman is 53 years of age; resides in an urban neighborhood; with 27 years of experience, earns LE 853 monthly, not married; with secondary or less education; poor health; does not benefit from health insurance; occupies a low professional rank (fourth or third); has one student in her household; her monthly expenses total LE 1494; doesn’t plan to work after retirement; and doesn’t know about early retirement incentives.
Results imply that region of residence does not influence the decision of retirement between the ages of 50 and 54, but residing in an urban district induces workers of both genders to retire more frequently between the ages 55 and 57, with a higher impact for women than men. Conversely, rural residents are more likely to retire at the age of 60. Age has the expected signs. Unsurprisingly, early retirement becomes less probable as the individual approaches the legal retirement age. In accordance with the predictions of the life-cycle model, the probability of early retirement decreases the further away from the retirement age a person is. That is, the older the worker gets the less likely he would retire early. For each additional year, the probability of both men and women, to retire at 50–54 decreases by two percentage points. It is also evident from the table that the probability of retiring at the old age increases by three percentage points for both genders.
The effect of experience on age of retirement is very small, close to zero, and has a negative effect for men and a mixed one for women. Women workers find their experience a determining factor in deciding when to retire. They seem to value their experience and prefer to stay longer on the job. That is, one additional year of experience to a female government sector worker, reduces her probability to retire at 50–54 age bracket by 0.4 percentage point, and has no significant effect afterwards. Unsurprisingly, being married does not enter as a factor in retirement decision for men retiring after age of 54. Married men are less likely to retire early, at 50–54. Being married however, does affect retirement decisions for women. They are less likely to retire at the legal age of retirement, but more likely to retire at the age of 55–57. The effect of being married on the retirement decision drops by 50% when it comes to preferring to retire between 50 and 54 years of age. This outcome is in contrast with that resulting from earlier studies on developed countries. Clark, (1988), and George, et al (1984) suggested men normally retire earlier than women, but in agreement with Ruhm (1996) who proposed otherwise, since women normally marry older men.
An interesting result is observed as regards to levels of education. In compliance with preceding descriptive statistics, education is not a significant factor in determining the retirement age for women, since close to two thirds of women sample are university graduates. In agreement with Quinn et al (1990), University graduates may have a greater investment in their human capital and therefore require longer stay on the job for a sufficient return on their investment. For men, non-university graduates were noticed to retire earlier than university graduates, especially between ages of 55 and 57. [End Page 92]
[End Page 93]
Expressing it differently, workers with low levels of education might face a relatively difficult financial situation than university graduates, that presses them to go into early retirement, hoping to rejoin the private sector; those with higher education probably do not face such a constraint (on average there is a difference of 55% between wages of workers of both levels of education), and probably have a higher preference for work relative to leisure. The effect of the perceived health status on the retirement decision operates in the anticipated direction. In general, and in accordance with Anderson and Burkhause, (1985); Colsher et al (1988); and Muller and Boaz (1988), good health standing is associated with the probability of staying longer in the labor market. Health effect reduces the probability of early retirement for both genders, and the effect for women is twice that of men. Following the same reasoning, benefiting from health insurance reduces the probability of a worker to opt for early retirement.
Occupational rank proved to be a significant predictor of age of retirement for women exclusively. Occupational rank reduces the probability of retiring between the ages of 55 and 57 for women. Women with higher ranks are more likely to retire late. The added worker effect has mixed results in this analysis. Working spouses have the opposite signs for men and women. The presence of a working wife in the household induces the husband to retire early. Yet, the presence of a working husband discourages women from retiring sooner. In other words, men with an active wife are expected to retire between the ages of 55 and 57. Women with an active spouse are more likely to retire late (at age of 60). This is somewhat a surprising result, since women are typically secondary wage earners in the household, and therefore, can afford early retirement; and contradicts Lilja’s (1996) findings in Finland that the presence of a retired worker in the house has a positive effect on early retirement and is gender neutral.
As previously mentioned in descriptive statistics, pension money is most probably used to help in kids education (over 50% of the respondents said they would use the pension money for their kids education and/or marriage expenses8). Hence, one would expect a higher number of students in the household would influence the early retirement decision negatively, especially when the worker is the sole earner. Results confirm that as the household embraces more and more school age kids, parents of both genders are less likely to retire early, confirming previous findings too (Perrachi and Welch (1994); and Reitzes et al (1998)). The wage effect is as anticipated. Since government wages are pre-determined according to the level of education, the effect of wage on early retirement is not significant. A finding that attracts attention is the effect of average expenditures in the household. Controlling for all covariates but monthly expenditures in the household seems not to affect the retirement decision for both genders.
Contrary to descriptive statistics--which showed findings without controlling for covariates, women at large were not planning to work after retirement. However, Results of
The preceding discussion suggests that early retirement decision is strongly determined by a myriad of factors. To visualize some of these findings, in what follows, [End Page 94] the patterns of the expected age of retirement along age and years of experience dimensions are presented graphically.
Breaking up the anticipated age bracket of early retirement into two categories, 50–54 and 55–57, Figure (4) reveals the most desirable age bracket of retirement, for both genders, is at 55–57. Rates of expected retirement at 55–57 ages for both men and women are generally high but exhibit no conclusive pattern with age for both genders. Again, women have a higher probability of retiring between the age of 55 and 57, than men do. The probability of retirement at 50–54 age brackets is comparable for both genders; starts up low and declines steeply thereafter as the person ages.
[End Page 95]
Figure (5) shows the predicted probability of each age category derived from the multinomial logit model. The model emphasizes the increased preference of women over men to retire at the age bracket of 55–57, and displays men exceeding women in their preference for retiring at the legal age. For a reference man, the predicted probability to retire at the age of 50–54 is 8%; the predicted probability to retire at the age of 55–57 is 58%, and 34% to retire at age of 60. For a reference woman, the relevant probabilities are: 10% and 68% and 22% respectively.
[End Page 96]
Figure (6) depicts these probabilities by years of experience, up to 40 years of experience. The higher years of experience the less likely women retire at the age bracket 50–54; this explains the downward, steeper curve for women. The effect for men is more or less constant and very low. The figure also reflects the opposite relationship for men vs. women when it comes to the probability of retiring between the age of 55 and 57 along years of experience. Additional years of experience raise the probabilities of retirement between 55 and 57 for women, whereas more years of experience, reduces the probabilities of retiring at the same age bracket for men. This conclusion may be a result of the fact that women are more educated than men and on average may have accumulated more savings over their work life, than men. Another way of looking at these results is to express the probability of retirement at certain age bracket as values of some explanatory variables are changed, while all other variables are kept fixed at their mean values. The following are simulations with the most relevant variables, presented in
Simulations produced in
The previous analysis is stretched to predict the number of employees retiring at different age categories, according to their type of work. Available data on the distribution of the government sector workers by age group, gender and type of activity for the fiscal year 2004/2005, Figure (7), reveals that the targeted group of workers (ages 51–60) represent 24% of all Administrative workers (Idary), 19% of Municipalities’ workers (Mahale-yat), and 27% of all Service Agency workers (Hay-aat). That is, on average, there is roughly one quarter of government sector workers facing the decision of early retirement. There are more men than women in each work activity, especially at municipality levels.
Predicting the probability of retirement by work activity is graphed in Figure (8). Administrative workers of both genders have the highest predicted level of early retirement: 78% for men and 83% for women, followed by service agencies (72% and 82% respectively), with women exceeding men in their preference for early retirement. Men working in municipalities are the least likely to retire early. All but 45% of them prefer old age retirement. Expressing these ratios in numbers,
[End Page 99]
Conclusion and Policy Implications
The purpose of this study is to develop and evaluate a model of employee retirement intentions. Retirement intentions are defined as intended retirement age and attitude after retirement. Such a model enables policy makers and others to identify characteristics of employees likely to support or resist early retirement. Besides, there are several advantages in knowing when most of the employees will retire and how they view their retirement. A predicted level of retirement increases the flexibility in personnel staffing and restructuring of the labor force. Among the most important statistics found in this sample data are: a little less than three quarters of the sample is expecting to retire on or at the age of 57. Women surpass men in preferring to retire between the ages of 55 and 57, while men exceed women in their preference to retire at the legal age of 60. Women excel over men in their educational status; therefore they hold higher positions and earn greater wages. In general, men are the head of their households and in many cases, the primary source of income, whereas women are the secondary earner in the household. Unlike men, women were found eager to learn new skills and take on special training when offered.
A sizable number of early retirees are planning to work after retirement, inevitably fighting over limited jobs with other unemployed and predictably raising the unemployment rate. What is more alarming is the fact that over half of those expecting to work after retirement, are planning to take on their own private venture. Adding lack of skills and the belief that no skills or experience are necessary for this risk may lead to partial or total failure, and eventually augmenting unemployment lines. It is also evident that early retirement schemes are not well announced ahead, and that only a small portion of respondents know about the policy. Estimating the multinomial logit model to examine the most important factors in the retirement decision, suggests that taking on early retirement is strongly determined by a myriad of factors. Results of the likelihood model showed that the probability of early retirement decreases the further away from the retirement age a worker is. The effect of years of experience is found negative, significant but very small and close to zero. Being married is confirmed to be a determining positive factor for women retiring early, and a negative one for men. The effect of health status is a determining factor in early retirement decision: good health standing is associated with the probability of staying longer in the labor market. Levels of education are not a significant predictor for women, but non university graduate men are more likely to retire [End Page 100] earlier than those with a university degree. This is a important result for policy purposes. If men are more likely to work after early retirement, and they are, on average, less educated than women, in addition to the fact that most of them are planning to take on their private venture, plus they do not think experience or special skills are necessary, then the probability of a successful venture becomes slim.
Occupational rank is an influential factor and reduces the probability of early retirement for women. Working spouses have the opposite effect for men and women. The presence of a working wife in the household induces the husband to retire early. Yet, the presence of a working husband discourages women from retiring early. An increasing number of school age students in the household has a negative effect on early retirement decision. Planning to work after retirement positively influences the early retirement decision for both genders. Prior knowledge of early retirement reforms has a favorable effect at early years of retirement for both genders but declines afterwards. The wage as well and expenditure levels do not have a determining effect on the early retirement decision for both genders.
Finally, simulation analysis proved that men without a university degree are 12 percentage points more likely to retire earlier than those with a university degree, who are 33 percentage points more likely to retire at the legal age of retirement.
The presence of a working wife increases the probability of early retirement for men by 8 percentage points, than those with an out of the labor force wife. Conversely, having a working husband in the household reduces the early retirement probability for women by 3 percentage points, but increases the likelihood of a legal age retirement by 26 percentage points, compared to women without working husbands. Women holding high professional ranks have an early retirement probability that is 10 percentage points lower than those with low professional rank, and 52 percentage points higher when it comes to retiring at the legal age. Finally, the effect of having three school age kids in the household has the same negative effect and magnitude for both genders, on early retirement, versus households without school age children.
This study is of great concern to policy makers for its contribution to the understanding of early retirement decision in Egypt. It is important when designing early retirement options in a way to avoid encouraging the most productive people to take advantage of them. Therefore, an understanding of these variables provides great insight towards policy decisions specific for its unique situations. It is also important for policy makers to predict acceptance rate of early retirement, the socio economic profile of the workforce as long as the most preferred retirement age.
The issue of early retirement needs careful consideration, as there is a recognized need for increased mobility in the labor force and for a shift in employment from the public to the private sector. This, combined with significant unemployment among the younger members of the labor force, implies the need for some public sector retrenchment. Some economists argue that it is undesirable that labor market problems be solved by generous early retirement provisions, which place a burden on the Social Insurance Funds and, at the same time, encourage workers to choose lower level of social protection. Others believe that offering early retirement is becoming an increasingly popular way to terminate the “lifetime” social contract between the government and the employees. [End Page 101]
Because normally older employees are paid more than younger employees are, and because older workers rate of pay may not be congruent with their current productivity, losing older workers through early retirement may save money. One important policy question is that of whether early retirement should be imposed in an ad hoc fashion on an ineffective older employee or whether it should be the option of any older employee to request early retirement under a plan that provides economic benefits that are fixed and known to all. If the latter, then care must be taken in the conception and formulation of the financial scheme, for it is likely that an institution would want the incentives to be the same for all. Moreover, if the firm’s most senior workers were at the stage of their life cycle in which wages exceed marginal productivities, the firm would best be managed by reducing their employment (i.e. older workers) rather than younger generation.
This assumption is realistic provided that one can expect that there will be stability in institutional policies as well as in the aggregate behavior of workers. For example, if unemployment, and the cost of living remain at high levels, continued employment after normal retirement age will be attractive and compulsory retirement will be resisted. Besides, the skepticism of early retirees about a better future may delay their acceptance for the current compensation packages. Add to that the fact that the slowing economy may deter potential interests.
Low average retirement ages, and markedly large shares of recipients getting retirement benefits may result in a large gap between the system dependency ration (ratio of recipients to contributors), and the demographic dependency ratio (ratio of old to young in the population). Although not yet exhausted, the Privatization Restructuring Fund has limited resources to speed up the program. This has limited the government’s flexibility for accelerating the program (a A detailed description is provided in CARANA, 2002a). To contain the risk of excessive disbursing and to minimize adverse selection, the best strategy is to identify the activities and the workers to be separated and to target the severance offer only to workers already identified as redundant (through benchmarking studies, for example, annual performance evaluation, combined with capping the salary of those who continue to perform unsatisfactorily, in addition to a well designed merit raise, rather than to all employees). Tailoring the severance package to worker characteristics (such as seniority or education) also helps induce the right self selection and contain costs, in a way to discourage the most productive workers to take advantage of them.
Early retirement plans may occur for different reasons and their announcement may provoke wide-ranging market reactions depending upon the institutional justification for the personnel reductions. It is clear that if a reason is provided it is either cost cutting in response to decline, or wider institutional restructuring. In cases of decline that continue to deteriorate, firms may not be able to recover. In this case, it is argued that the faster the firm moves toward financial losses, announcements of continuous cost-cutting schemes such as scaling down may signal the market that it is still in decline and may still be in a worse condition in the future. On the contrary, early retirement announcements not declared or publicized during a decline may indicate a firm adapting to a new situation rather than just a reduction in its labor force, and elicit workers to take advantage of them. [End Page 102]
Finally, as this study has provided the rationale for, it is very important for the government to provide training workshops for early retirees on how to best invest their pension money, and open offices of help and support for these projects. In the Egyptian social contract, government, public sectors, and SOE workers are guaranteed lifetime employment and benefits. When preparing for labor restructuring encompasses high levels of surplus labor and when social safety nets and unemployment insurance in labor laws are inadequate or lacking, the government should be involved in the support and funding of special programs to deal with inexperienced small business owners and unemployment problems.
1. Kikeri (1999) reports that payment ranged from 18 months salary in the case of Brazil to two and three years salary in restructuring programs in Argentina and Bangladesh.
2. It is well agreed upon that the age of 45 is the age at which worker’s maximum productivity is achieved, according to Marchetti (2002).
3. For a complete description of survey instruments and procedures, refer to: Ramadan, Mohamed (2006). “Analysis of Early Retirement Survey”, in Arabic; IDSC, January 2006.
4. The original sample has 3437 observations, of which 40 observations had incomplete data points. The variable “Having a working spouse” had 404 missing observations, which were imputed using logit model. (see van Buuren et al. (1999) for further technical details.
5. For a complete descriptive statistics of the sample, refer to: Ramadan, Mohamed (2006). “Analysis of Early Retirement Survey”, in Arabic; IDSC, January 2006.
6. Illiterates are those with less than 4 years of education working in supportive jobs, according to CAPMAS (Central Agency for Public Mobilization and Statistics) definitions.
7. Sometimes the respondent refers to his professional rank as an equivalent to what he is supposed to be at, not necessarily the current rank.
8. The survey doesn’t provide additional information on the age or educational level of these students.
9. Since the predicted probabilities are drawn from the survey of workers 50–57 years of age; and the available data reports number of workers in the age category of 51–60.