Chapter 1 The Discrepancy between the Desire of Working and the Reality for the Elderly

Transcription

1 Chapter 1 The Discrepancy between the Desire of Working and the Reality for the Elderly Tachibanaki Toshiaki and Urakawa Kunio 1. Introduction Japan faces currently an ageing trend, and thus it provides Japan with decreases in labour force and economic prosperity. One of the policy tools to stop such negative effects is to increase the labour supply of the elderly, and thus to utilize the work-force of the elderly more efficiently. Under such conditions economists are interested in investigating the working behaviour of the elderly with an attention to the effect of wages and some other labour-related variables including the effect of social insurance policies. We can raise the studies like Seike(1993), Abe(1998), Yamada(2000), Mitani(2001), Higuchi and Yamamoto(2002), Seike and Yamada(2004), Higuchi, Kurosawa, Sakai, Sato and Takeishi(2006), and some others. (Table1) Availability of individual survey data enabled these authors to investigate various subjects econometrically. The main concerns of these studies are the effect of health, work history, non-labour incomes of the elderly, and their family conditions, employment conditions of local areas, employment policy of both firms and government, etc. An attention has been paid also to some other variables such as wages, human capital of workers, and career development policies at firms. Nearly all studies mentioned above investigated the actual effects of these variables on the working behaviour of the elderly. It is possible to anticipate the fact that the current working conditions would be different from the desire of the elderly. Several examples can show the discrepancy between the actual story and the desire. Some elderly are now unemployed, even though they desired to work somewhere initially. Although they hoped to work as part-time employees, they are self-employed currently. There should be a number of the elderly who engage in the different working positions currently from their initial desires. If the degree of the discrepancy were large, the work motivations of these elderly would be 1

2 lower. Thus, it is desirable to prepare working positions for the elderly possibly which are close to their initial desires because the discrepancy implies that human resources of the elderly are not fully utilized, and thus that human resource allocations are not optimum. It would be a useful subject to inquire the reasons why such discrepancy has occurred, and to seek policy options in order to reduce such discrepancy. Shimono and Tachibanaki(1984), Tachibanaki and Shimono(1985, 1994) investigated such discrepancy seriously, and obtained the following result; the choice of the actual state is based on various economic conditions and individual labour force conditions more strongly than the preference of the desired state. Also, it showed some significant effects and some other non-significant effects regarding the impact of various variables on the determination of labour supply behaviours of the elderly. The data used by them, however, are old because it was a story in the 1980s. It should be interesting to study the current story in view of the fact that no studies have been made recently regarding this subject. Another important reason why the current study is useful is that there were several policy reforms which would have changed the labour behaviour of the elderly, such as a change in the amount of pension benefits when an elder person works, and a reform in the employment law for the elderly, etc. (Table.2) It is a worthwhile attempt to know the effect of these changes and reforms on the working behaviour of the elderly. 2. Working Activity of the elderly This part is very brief because we are able to refer to a large number of documents and studies on working activities of the elderly in Japan. We write here only very important observations which led us to conduct this research. First, the rate of labour force participation for the elderly in Japan has been higher considerably in comparison with the OECD countries. (Figure.1) Second, the main reason why the elderly in Japan desire to work, and work in fact is that a large number of the elderly find that they do not have enough amount of income to live during their retirement period. We raise Seike and Yamada(2004) for it, and it is worthwhile to mention that this is true, in particular, for the elderly with lower incomes. Third, it is emphasized that a non-negligible number of the elderly give up working for 2

3 various reasons in spite of the fact that they desire to work even if they are old. We know, for example, a large number of the elderly are unable to find jobs. There are numerous reasons why the elderly cannot satisfy their desire of working. (Figure.2) The purpose of this paper is to present the reasons and seek policy suggestions. 3. The Desire and the Actual State for the Working Behaviour of the elderly. 3.1 Data The data used in this study is Survey on Jobs and Living Conditions for Employees who passed their Statutory Retirements Ages, 2002, provided by the Agency for Employment Helps for the Elderly and the Handicapped. Specifically, it asks male people whose ages are years old, and who passed statutory retirement ages after working at private companies. It also includes some male people who separated from the previous employers even before the age of 50s. The number of questionnaires is 33,000, and 3732 observations are available. The response rate is 11.3%. It asks various questions on the working conditions like labour force status, income and work history before the retirement age, and various questions on living conditions. Table 3 presents the summary statistics of the data, including age, labour force status, marriage, education, residential location, income. One important characteristics derived from this data source is that the rate of self-employed among total labour forces is 6.6%, which is lower than the national figures. Thus, it would be preferable to describe that we intend to study the working and living conditions of the employed elderly mainly. 3.2 The Discrepancy between the Desire and the Actual Story on Labour Force Status It is well-known that the Japanese elderly people desire to work as possible as they can and they show, in fact, a high rate of the labour force participation rate. Even a large number of the elderly, who are currently not-in labour force, show their desire to work. Table 4 presents the gap between the desired retirement age and the actual retirement age. It shows the fact that a less than fifty percent of the already retired elderly express that they retired at their desired age of retirement, implying that they wanted to postpone their retirement age. They had to retire earlier than their desired retirement ages. For example, 3

4 about 40 percent of the elderly who retired at the age 60 say that their desired retirement age is older than the age 61. About 30 percent of the elderly who retired at the age 65 declare that they wanted to retire after the age 65, although about 10 percent declare that they wanted to retire before the age 64. As for the elderly who say that they are temporally away from working, not-working for unclear reasons, desiring to work if some favourable conditions appear among the already retired elderly, or seeking jobs seriously among the currently unemployed elderly, even a high rate of people declare that they would retire after the age 65. This high rate, namely 30 percent of the late retirement (say the age 70), is also observed in Korea, although figures in the U.S., Germany and France show that only 15 percent hope to retire at the age 70. In sum, there are numerous elderly who cannot find jobs, or who were obliged to retire earlier than their desired ages, even though they wish to work. Next, an attention is paid to the difference between the desired occupation (or working status) and the actual occupation. The survey gives the following five choices for the desired status: (1) self-employed, (2) full-time employee, (3) part-time employee, (4) NPO or voluntary, and (5) desiring retirement, and gives the following four choices for the actual one: (1) self-employed, (2) full-time employee, (3) part-time employee, and (4) non-working. The figure in this Table 5-1, 5-2, 5-3 signifies the distribution rates of the actual choices within each desired choice among five desired occupations (or working status). Table 5-1 suggests that the coincidence rate, which signifies that the elderly are able to occupy working status the same as the desired one, is about 50 percent. The elderly whose desired occupation (i.e., part-time employee) accounts for about one-fifth among the total elderly, but about only 40 percent of these people are able to find their part-time jobs. About 50% are non-working. Another desired occupation (i.e., self-employed) is given by about 10 percent among the total elderly, and only about 40 percent are able to find their self-employed jobs. The remaining figures are 45 percent as non-working, and 10 percent are employed workers rather than self-employed workers. The similar result is observed for the elderly who desire to work as full-time employees. In sum, there are a large number of the elderly who cannot find their working status which is the same as their desired status. One quick note is added. The survey asks a desired occupation (or working status) not for 4

5 a current desire but for a desire during the elderly age in general. Thus, it is somewhat risky, strictly speaking, to compare between desired occupations and actual occupations. This risk, nevertheless, does not deny the conclusion at all that there is a considerable degree of the discrepancy between desired occupations (or working status) and actual occupations (or working status). Table 5-2 and 5-3 show the similar distributions with table 5-1, which are classified by age and education, respectively. Table 5-2 presents the fact that the age difference, namely between under the age 64 and over the age 65, is clear. As for full-time employees, the coincidence rate for under the age 64 is 53.9 percent, while it is 45.5 percent for over the age 65. The difference is about 8 percent point. As for self-employed it is 34.4 percent for under the age 64, while it is 46.7 percent for over the age 65. The difference is about 13 percent point. One important difference between under the age 64 and over the age 65 is seen by the opposite direction concerning the desired rate and the actual rate between the elderly who desire to be full-time employees and the ones who desire to be self-employed. The similar result is obtained also for education based on Table 5-3. It is found that the coincidence rate, 54.3 percent for university and graduate school graduates is higher than that for junior and senior high graduates and junior college graduates, 44.7 percent for the elderly who desired to work as full-time employees. It is, however, different in the case of self-employed desire because the coincidence rate for junior high graduates and etc., 49.3 percent is higher than that for university graduates and etc., 36.5 percent. Table 6 is different from Table 5 because the base is the actual state rather than the desired state in Table 5. In other words, the distributions based on the desired state for each actual state are the main concerns. The similar result with Table 5 is obtained regarding the difference between the desired state and the actual state. An interesting result, nevertheless, appears for the elderly who are non-working currently because about 60 percent declared their desire for retirement and the remaining 40 percent desired to work. It was revealed previously in Table 5-1 that over 90 percent of the elderly who desired to retire were, in fact, non-working. Finally, it is noted that about 50 percent of the elderly, who desire to work among the currently non-working elderly, desire to work as part-time employees. 5

6 3.3 Distributions of Other Variables Regarding Desired and Actual Status This part uses tables which integrate elements from the five classifications into four classifications in the case of desired status. In other words, (3) part-time working, and (4) NPO and voluntary, are combined into one element. Table 7 presents the distributions based on average age, health status, education, work history, residential location, non-labour income for each categorie. Table 7 indicates several useful results. First, we are concerned with the left-hand side, namely the actual state. As for full-time employees, the rate of healthy people is 87.9 percent, the rate of university and graduate school graduates is 61.1 percent, the rate of the elderly who worked at large firms (i.e., over 1,000 employees) is 59,2 percent. The result, however, for non-working elderly shows that the rate of healthy people is 76.5 percent which is lower than the average figure of 79.6 percent. Nevertheless, annual public pension payment is 2,765,400 yen which is slightly higher than the average figure of 2,676,400 yen. We are concerned, next, with the right-hand side, namely the desired state in Table 7. The same result with the one for the actual state is observed for the desired state. Concretely speaking, as for full-time employees, the rate of healthy people, and the rate of university and graduate school graduates are high. Also, the elderly who continued to work in the former company after the statutory retirement age desire to work with a high rate as full-time employees. Concerning the elderly who desire to be self-employed the rate observed for the elderly who worked at small firms (i.e., less than 30 employees) is 12.7 percent which is slightly higher than that for other workers. The rate of healthy people for the elderly who desire to retire is 72.6 percent, which is slightly lower than the average 80.0 percent. Also, the public pension payment is slightly higher than the average figure of 2,791,000 yen. The overall result explained so far suggested that age, education, health, work-history before statutory retirement age, non-labour income and some other variable have impacts on the probability whether the elderly are able to achieve their working statuses which are the same as their desired status. We examined these impacts based on the descriptive statistics. Our next task is to apply some quantitative methods to obtain more reliable results. 4. Working Behaviours and Wages of the Elderly 6

7 4.1 Econometric Method (I) The probit model is applied for the analysis of the effect of various variables on both actual and desired status, after controlling for the other characteristic variables of the elderly. The model is written as follows, P( y r = 1 X, Z) = G( α + X β + Z γ ) (1) P( y h = 1 X, Z) = G( α + X β + Z γ ) (2) The dependent variables in these equations take unity if the elderly work, and zero otherwise. Of course, y h indicates desired state, while y r indicates actual state. We consider the following independent variables for the vector X ; a dummy variable for the age 64, a dummy for health, a dummy for the necessity of caring some family members, a dummy for education, a dummy for firm size, a dummy for executives, a variable for employment help at previous employers (i.e., firms at statutory retirement ages), a variable for skill development at previous employers, a variable for skill development at previous employers, a variable for the estimated decreasing rate in wages, a dummy for residential area (i.e., Kanto area.) We consider the following independent variables for the vector Z which consists of various economic variables. They indicate the elderlys economic conditions such as public pension payment, non-working income, and a dummy for housing loan. For the variable for pensions, just as in the analysis of Ogawa (1998), the original public pension, i.e. the amount received when no deductions are made, has been used. When an individual decides whether to be employed, the amount of pension that the person considers should normally be the original amount as defined here, rather than the amount actually received, which might have been reduced from the original. The original pension amount is independent of the work status. Therefore, the problem of simultaneous decision bias related to the pension amount and work status no longer exists if the original pension is used. The original amount of a public pension has been determined by applying the responses to the question on the rate of deduction from the old-age pension for active employees to the following equation: For respondents who presented a 100% deduction in their answers, the educational background of the recipient of the employee pension without deductions and the average 7

8 pension by work experience were applied to calculate the original amount of the public pension. Figure 3 depicts the relationships between the amount of public pension that was to be received originally and the employment rate, or the rate of individuals wishing to work. This figure suggests some interesting points. First, those with an original amount of public pension less than 80,000 yen display a substantially high rate of employment at 86.3%. Not surprisingly, economic conditions have considerable effects on employment. Meanwhile, the percentage of respondents who answered wishing to work in some way proved to be slightly lower than the actual employment rate, which was 80.3%. Secondly, although the actual employment rate and the rate of those wishing to work decrease as the amount of public pension increases, those who receive a pension of 280,000 yen or more maintain a high employment rate of 44.6%, which nearly equals the employment rate of those who receive between 80,000 and 160,000 yen in pension. The third point is that, overall, the actual employment rate and the rate of those wishing to work deviate considerably at the middle level with pensions between 160,000 and 280,000 yen. Although this group of individuals is considered to be free of the necessity to work merely for living, more than half retain a desire to work. However, the actual employment rate is limited to slightly more than 30%. Table 8 explains in what way variables for employment help, for skill development, and for the estimated decreasing rate in wages, etc. are constructed. 4.2 Estimated Results(I) Labour Supply Functions Table 9-1 presents the estimated results for both equations (1) and (2). The result for equation (1) provides us with the nearly consistent results with the ones obtained by Higuchi and Yamamoto (2002), Seike and Yamada(2004), and some other studies. Since we have a consensus in this field, we write here only the conclusion. If the elderly are healthy and were executives, they tend to work with a higher probability, while if they receive a high amount of both public pension benefits and non-working incomes, they tend to work with a lower probability. If the elderly worked at smaller firms (i.e., less than 30 employees), their 8

9 probability of working would be higher than those at employees firms. The effect of the estimated decreasing rate in wages is negative, implying that they do not work if their wages decline a lot. The final observation with respect to wages is important because the behaviour of the elderly is influenced by wage payments fairly significantly. We found a new and useful result concerning the effect of education and employment helps by previous employers. As for education variable we obtained the result such that junior high graduates show a negative effect, while university and graduate school graduates show no obvious effect when we take senior high and two-years college graduates as a reference group. The variables of both employment helps and skill development by previous employers did not produce any significant effect. It is noted, however, that an employment help variable showed the positive effect with the 1 percent significance level for the equation (3) where the elderly worked or had an experience of working after the statutory retirement age, although skill development variable is not statistically significant. The result is shown in table 9-2. Incidentally, the average retirement age in this sample is 59.8, and the rate of retirement for the elderly whose age was younger than 60 is 90.9 percent. P( y a = 1 X, Z) = G( α + X β + Z γ ) (3) where the meaning of the dependent variable y a was already explained. Let us examine equation(2) where a desired state is asked. The estimated result is somewhat different from equation (1) where an actual state is asked. A dummy for the necessity of caring some family members is positive with statistical significance, while it had no effect for the determination of the actual state. It is possible to anticipate that the requirement of more incomes would be higher if one or two family members need some caring help. The actual story, however, prohibits from working because the elderly face a strong barrier against working because caring is a demanding and time-consuming task. Anyway, the consideration of a dummy variable for caring-necessity gives us an important factor to investigate the discrepancy between actual status and desired status concerning the working of the elderly. Education provides us with an interesting outcome. It was described previously that a dummy for junior high school graduates was negative on the both effect of the actual state and desired state. 9

10 Those results, probably, suggest that junior high graduates initially hoped to retire earlier. The different story appears for university and graduate school graduates because they show their willingness to work in view of the positive effect with the 5 percent significance level. The effect, however, on the actual state was not clear. They hope to work fairly strongly. They do not find, however, jobs so easily. We summarize the empirical findings here based on Table 9-1 and Table 9-2. First, our data and the other popular data called, Survey Working Status for the Elderly published by the Ministry of Labour and Welfare indicate that various variables like health status, public pension, non-labour income and experience as an executive, have significant effects on the elderlys working behaviour. Another useful finding is that a high degree of decreasing rate in wages encourages the elderly to retire from the labour market. In the middle of the 1990s the increasing rate of wages for generations of the age years old was fairly steep in Japan according to the OECD(1998) Employment Outlook. It showed, however, a considerable degree of declining rate in the age years old. This acute decrease in wage payments encourage to the elderly to retire. The Japanese data, namely the Ministry of Labour and Welfare s Wage Structure Survey indicated the following observations; the highest wage figure in 1990 was in comparison with the age 20 s figure of 100. It declined to in 2004, implying that there was a considerable decrease (about 70 percent) in the wage curve according to ages. In other words, the age-wage profiles tends to be a more flat nature. Yamada(2000) also refers to this fact that this is likely to encourage the elderly to retire fairly strongly. Second, it was true that employment help policies such as prolonging the retirement age, continuing employment after the statutory retirement age, and introducing other employment opportunities after retirement from firms were effective at least temporarily. The recent data, however, do not support the above finding unfortunately. This is true for skill development policies by previous employers. These results are observed at least for the elderly whose ages are years old. We are required to take other or different forms of policies in order to encourage the elderly to work. Third, variables like education and necessity of caring some family members have the different impacts on actual states and desired states regarding the elderlys working 10

11 behaviour, although some other variables do show the common (i.e., positive or negative) effects between actual states and desired states. In particular, the behaviours of junior high school graduates and of university and graduate school graduates are considerably different Estimated Wage Funtions We estimate wage equations of the elderly, based on both the quantile estimation method and the Heckman s two-step method. The first-step dependent variable is at equation (1), and its independent variables are the same as equation (1) except for the estimated decreasing rate in wages. Table 10 shows the estimated result. The estimation procedure based on the quantile estimation method is given by optimization problem in equation (5) under the regression equation (4). Chamberlain(1994) is a representative example which was estimated by this method, and a useful survey article was provided by Koenker and Hallock(2001). ln w = α + X β + Z γ (4) i q ' i q ' i q y r Min K N β R i: y xβ i q ln w i i α q X β Z γ ' i q ' i q N ' ' + (1 q) ln wi α q X iβ q Z iγ q ' i: ypxβ i (5) where ln w is the log-form of per-hour wage figures, and it is given by monthly wages and weakly working hours. We know that the usual OLS estimation is made by minimizing the sum of residuals square without giving any weights. Thus, we are able to know only the average story. The quantile method, however, gives some weights on each quantile. In other words, we minimize the sum of weighted residuals square, and thus we are able to obtain the conditional distribution of the dependant variable (i.e., 100q % quantile). In other words, we are able to estimate the effect of each independent variable on wages for each quantile income class separately. Table 10 is the estimated result. We examine, first, the result based on the Heckman s two-step method which tried to take out a sample selection bias. The result suggests the following observations. The positive effects on wages are provided by the variables like a dummy variable for under the age 64, a dummy for university graduates, a dummy for larger firms, a dummy for executives, a dummy for continuing employment after the statutory 11

12 retirement age, and a dummy for introducing other employment opportunities. Several variables are related to so-called human capital variables. It is, nevertheless, unclear whether the elderly who accumulated their human capital well are likely to receive helps from previous employers more easily, or the relationship between a parent company and his subsidiaries can encourage employment opportunity of the elderly. These factors are important to support employment opportunities of the elderlys, and thus we need a further study. Second, the result based on the quantile method is discussed. The most valuable finding derived from a new estimation method appears in the different impacts on wages by wage classes. Let us show such difference. The result for the highest wage class (i.e,. q=0.9) indicates that a dummy for junior high graduates is negative statistically, while it is statistically insignificantly for university and graduate school graduates. Some other variables such as a dummy for larger firms, a dummy for executives, a dummy for introducing other employment opportunities by previous employers, and Kanto area are all significant positively. Only one variable, namely continuing employment after the statutory retirement age, is statistically insignificant. A considerably different outcome appeared for the lowest decile class (i.e., q=0.1). Although the result is the same as for the highest decile class, namely a dummy for larger firms and a dummy for executives are positive with statistical significance, a dummy for continuing employment after the statutory retirement age and a dummy for university graduates are positive with statistical significance. A dummy for junior high school, a dummy for introducing other employment opportunities, a dummy for Kanto area are not statistically significantly. In sum, a large number of the signs of the estimated coefficients are opposite between the highest wage class and the lowest class. The result implies that each effect works differently by the wage class, and at the same time we have to take into account these different effects when we formulate policy suggestions to improve working conditions of the elderly. In particular, we have to be careful about the actual implementation of introducing other employment opportunities, and of continuing employment after the statutory retirement age. Incidentally, the effect of skill development on wages is always statistically insignificant. The 12

13 above result was obtained for samples who are currently working. We did the similar job for samples who once worked, but currently retired. The empirical result for wage equations are almost the same. Thus, no detailed explanation is provided for these samples. 5. Discrepancies between Actual States and Desired States regarding Working Behaviours 5.1 Econometric Model (II) This part pays attention to the elderly who are currently non-working. At the same time, we are concerned whether those people desire to work if some opportunities are available. The main econometric method is the multinomial logit model, controlling for several variables. The estimated equation is given by (6) where ln( p p 1 m / p1) = α m1 + X β m1 + Z γ m1 (6) signifies retirement status, and also no desire for working. Thus, this is a reference group for the other choices. There are three choices for the elderly who desire to work, if some good conditions are available, (1) self-employed, (2) full-time employees, and (3) part-time employees. We intend to estimate the effect of each independent variable when some differences in probability appear between p1 and the other choices. Vector X and Z are the same as the previous models. 5.2 Estimated Results (II) Table 11 presents the estimated result for each choice, namely (1) self-employed, (2) full-employees, and (3) part-time employees in relation to the base choice of retirement and no desire for working. Interpretations for the effect of each independent variable are provided based on this table. First, a dummy variable for age under 64 years old is positive with statistical significance for (1) self-employed and (3) part-time employees. As for (2) full-time employees age is not statistically significant. It is possible, nevertheless, to point out that the degree of desiring works is higher for the age class of earlier 60s than that of late 60s. We are concerned with the effect of health status, and necessity of caring some family members. We find the positive effects both for (3) part-time employees by the former (i.e., 13

14 health), and for (1) self-employed by the latter (i.e., caring). It is quite reasonable to understand the fact that the elderly take into account family conditions in the determination of working status, and that they hope to be self-employed, if they have family members who should be taken care. One worthwhile observation regarding self-employed is that only half were able to find such opportunities as self-employed. It is reminded that the elderly in this statistical source were once working around the statutory retirement age. Therefore, the great majority of the samples have no experiences as self-employed. This condition gives hard time to find self-employed jobs for these people. Some policy helps would be necessary. Next, we pay attention to the effect of education. A dummy variable for university and graduate school graduates is positive with statistical significance for (1) self-employed and (2) full-time employees. As table 4-3 showed, university graduates expressed their desire of working (about 60 percent) among the elderly who are currently non-working. Combining this result with the finding in Talbe 11, it is possible to conclude that the highly educated elderly keep fairly high working motivations, although their labour force participation is still higher in the OECD countries. It is very likely that we are losing highly qualified human resources, if these elder people did not find any suitable jobs. We are concerned with the effect of firm size of previous employers. The coefficient for smaller firms (i.e., less than 30 employees) on (2) full-time employees is positive with statistical significance, while the one for larger firms (i.e., over 1,000 employees) on (3) part-time is also positive. The elderly who worked at larger firms desire to work with more flexible working hours, while those at smaller firms desire to work as full-time employees. The difference between those at larger firms and those at smaller firms arises from the difference in saving figures. Our statistical source provided us with the amount of retirement allowances. It is million yen for larger firms (i.e., over 1,000 employees), while it is 9.16 million yen for smaller firms (i.e., less than 30 employees). The difference is quite large. It is responsible to explain why the different approach for working status is adopted between these two different groups. Let us examine the effect of a dummy for executives. It is statistically insignificant on both (2) full-time employees and (3) part-time employees. 14

15 Finally, we examine the impact of the elderlys economic conditions. Variables like a dummy for public pension and the amount of non-labour income are negative with statistical significance on (1) self-employed and (3) part-time employees. The result is quite reasonable because the elderly who have a large amount of economic resources by public pension and non-working income do not have strong incentives of working. A dummy for housing, nevertheless, shows a different story, judging from the statistically significant positive effects on these three choices. In other words, the elder who have to repay some the amount of housing loans desire to work. 6. Conclusion This paper examined the discrepancy between desired states and actual states regarding the elderlys work behaviours, and found that there is a considerable degree of the difference between the two states. For example, about only half of elderly people could find their part-time jobs among the elderly who desired to be part-time employees, and 40 percent did not find any jobs. The similar result was obtained for the elderly who desired to be self-employed. Another example is that there are a large number of the educated elderly who are currently non-working, but cannot find any full-time jobs or work as self-employed. One important fact finding in this work is that under-utilization of educated and skilled human resources in labour force participation is observed in Japan in spite of the fact that a higher rate of labour force participation is seen among the elderly compared with the other OECD countries. There are so many reasons why the elderly are unable to find their appropriate jobs despite they desire to work in various forms. It is crucial to prepare various institutional reforms and helps for these elderly people in order that Japan can utilize their human resources more efficiently. This is important because Japan will face a labour shortage soon due to the current ageing trend. One candidate of the policies is to decrease the wage-profile of employees where ages are 40s or 50s years old (near the retirement age) slightly in order to avoid a drastic decrease in wage payments after the statutory retirement age because we found that the elderly were likely to lose work motivations if they encountered a drastic wage decrease. Additionally, one example is to help the elderly to engage in self-employed jobs because 15

16 many elderly desire to work as self-employed after the statutory retirement age. The role of public sectors and the related institutions is crucial. Another example is that the elderly who give up their working activity if some family members need caring. Reforms in the social security systems are necessary. Fortunately, the current Japanese elderly desire to work strongly, and their motivations are quite high. It is necessary to prepare the system which enables them to find appropriate jobs and to work more efficiently. The system, which should be modified, includes hiring rule, working conditions such as working hours and wages, social security system, and some others. The past Japanese system was constructed and worked well under the condition that the population growth rate was positive. The current ageing trend, however, urges to modify the system, as Seike and Yamada(2004) emphasized. In particular, a drastic changes for the labour market for the elderly are required, and the government, employers and workers are responsible to make such changes in order to utilize human resources of the elderly more efficiently. 16

17 Reference Abe, Y.(1998), Labour Supply and Old-Age Pension for Active Male Employees in the 90s, Nihon Keizai Kenkyu, No.36, pp Japan Center for Economic Research. (in Japanese) Amemiya, T. and Shimono, K. (1989), An Application of nested logit models to the labor supply of the elderly, The Economic Studies Quarterly, Vol.40, pp Higuchi, Y. and Yamamoto, I.(2002), Labour Supply Behaviours of the Male Elderly in Japan. Kinyu Kenkyu, , pp (in Japanese) Higuchi, Kurosawa, Ishii and Matsuura(2006), An analysis of Social Security Reforms on Labour Supply Behaviour of the Male, RIETI Discussion Paper Series, 06-J-033, pp.1-pp.32. (in Japanese) Higuchi, Kurosawa, Sakai, Sato and Takeishi(2006), The Effect of Caring on Working Behaviours of the Elderly, RIETI Discussion Paper Series, 06-J-036, pp.1-pp.30.(in Japanese) Mitani, N.(1997), Labour Market and Intra-Firm Wage System, Keiso Press. (in Japanese) Mitani, N.(2001), Employment Policies for the Elderly and Labour Demand, Inoki, T. and Otake, F. eds, Economic Analysis of Employment Policies, Ch.11, University of Tokyo Press, p (in Japanese) Oishi, A. and Oshio, T.(2000), Retirement behaviour and Social Security Assets of the elderly, The Japanese Journal of Social Security, Vol.35, No.4, National Institute of Population and Social Society Research, pp (in Japanese) Seike, A.(1993), Labour Market of Aging Society, Toyo Keizai. (in Japanese) Seike, A. and Yamada, A.(1998), An Hazard Analysis of Social Security and Employment System on Retirement Decidion of the Elderly, Mita Syogaku Kenkyu, Vol.41, No.4,pp (in Japanese) Seike, A. and Yamada, A.(2004), Economics of Elderly s Working Behaviour, Nikkei Inc. (in Japanese) Shimono, K. and Tachibanaki, T. (1984), An analysis of the working behaviour of the elderly: a comparing men and women, Quarterly of Social Security Research, Vol.19, Spring, pp (in Japanese) 17

18 Tachibanaki, T. and Shimono, K. (1985), Labor Supply of the Elderly, Economic Review, Vol.36, No.3, July, pp (in Japanese) Tachibanaki, T. and Shimono, K.(1994), Private Savings and Lifecycle, Nikkei Inc. (in Japanese) Yashiro, N., Oishi, A. and Futagami, K.(1995), The Factor of the Working Behaviour of the Elderly, Research on Utilization of Human Resources and Changes of Occupational Structurs, Japan Center for Economic Research pp (in Japanese) Yamada, A.(2000), Wage Decreasing and Income Security in Retirement Process, The Japanese Journal of Social Security, Vol.35, No.4, pp (in Japanese) Becker, G., (1975), Human Capital, NBER Carmichael, L. H. (1989), Self-Enforcing Contracts, Shirking, and Life Cycle Incentive, Journal of Economic Perspectives, Vol.3, No.4, pp Chamberlain, G. (1994), Quantile Regression, Censoring and the Structure of Wages, in Advances in Econometrics. Christopher, S, ed. New York: Elsevier, pp Duval, R. (2003), The Retirement Effects of Old-Age Pension Systems and Other Social Transfer Programmes in OECD Countries, OECD Economic Department Working Papers, No.370, OECD, Paris. Fields, G. S. and Mitchell, O. S. (1984), Retirement, Pensions, and Social Security, Mit Press. Koenker, R. and Hallock, K. F. (2001), Quantile Regression, Journal of Economic Perspectives, Vol.15, No.4, pp Lazear, E. P. (1979), Why is There Mandatory Retirement? Journal of Political Economy, Vol.87, No.6, pp OECD(2005), Employment Outlook 18

19 Table 1 Early Researches on Labor Supply of the Elderly Data Econometric Approach Classification of Job Status Arrangement for simultaneous dicision-bias of public pension Seike Yamada(2003) The Employment Status Survey of the Elderly (2000) Probit Permanent employment Casual employment Disemployment Non-work Using dummy-variable for qualification as a recipient Higuchi Yamamoto(2002) Mitani(2001) The Employment Status Survey of the Elderly (1992, 1996, 2000) The Employment Status Survey of the Elderly (1996) Multi-logit Nested-logit Multi-logit Permanent employment Casual employment Disemployment Non-work Continued employment Job change Disemployment Non-work Employee Self-employed Non-work Using estimated actural public pension benefit / Oishi Oshio(2000) Abe(1998) Ogawa(1998) The Employment Status Survey of the Elderly (1996) The Employment Status Survey of the Elderly (1983, 1988, 1992) The Employment Status Survey of the Elderly (1983, 1988, 1992) Probit Probit Probit Retirement Non-retirement Work Non-work Work Non-work Using estimated actural public pension benefit Using estimated actural public pension benefit Tachibanaki Shimono(1994) The Employment Status Survey of the Elderly (1983) Multi-logit Pert-time employee Fulltime employee Selfemployed Non-work / Amemiya Shimono(1989) The Employment Status Survey of the Elderly (1980) Nested-logit Pert-time employee Fulltime employee Selfemployed Non-work / Resorce)Addition to Higuchi Yamamoto(2002) 19

20 Table 2 Policies to create emloyments for older people Year Measures, etc 1971 Law concerning Stabilization of Employment of Older Persons Health and Medical Service Law for the Elderly amendment (health services facillities for the elderly) Law concerning Stabilization of Employment of Older Persons (Employers are obliged to endeavor to introduce the 60-year-old retirement age) Pension system reform (Reform for the Old-Age Employees' Pension for people in their early 60s) Pension system reform (adjustment of total benefit amount growth, etc) Social Welfare Law 2004 The Law Concerning Stabilization of Employment of Older Persons. The Law aims to (1) introduce an obligation for employers who have mandatory retirement system to take one of the following steps, beginning in April 2006, (2) improve the measures to promote reemployment of middle-aged and older people, and (3) enhance the measures to ensure temporary and short-term employment opportunities for retired people Pension system reform (introduction of mechanisms of fixing upper limit of premiums and of automatic adjustment of benefit levels; establishment of the Pension Reserve Fund Administration and Investment Agency, etc.) Source) Ministry of Health, Labour and Welfare, Annual Report on Health, Labour and Welfare , Ensuring Social Security in Cooperation with Local Actors. 20

21 [Figure 1] A Change in the Employment Ratio (male between aged 55 and 64) 90 Japan 80 Sweden (%) United States EU(15 countries) EU(15 contries) France Dutch Sweden United Kingdom United States Japan Resources) EU, Employment in Europe 2005, OECD, Employment Outlook 21

22 Figure 2-1 Reason for not-working currently [Male] Other, 8.4% Health problem (Family member), 5.3% Health problem (individual), 28.8% Mismatch of occupational category and desire, 16.8% Mismatch of labour hours and desire, 5.2% Mismatch of wage and desire, 2.6% Mismatch of other conditions and desire, 4.9% No work though not stuck to the condition, 27.9% Resource) Health, Labour and Welfare Ministry, Survey on Employment Conditions of Elderly Persons 2004 Table 2-2 Reason for not-working currently [Female] Household duties, 11.7% Health problem (Family member), 10.2% Other, 6.3% Mismatch of occupational category and desire, 10.2% Mismatch of labour hours and desire, 7.4% Mismatch of wage and desire, 0.5% Mismatch of other conditions and desire, 7.5% Health problem (individual), 22.8% No work though not stuck to the condition, 23.3% Resource) Health, Labour and Welfare Ministry, Survey on Employment Conditions of Elderly Persons

23 Table.3 Summary Table of Descriptive Statistics N Age Job type Consort relation Academic Background Regional block Earned income+ Public pension fraction Below the age of % Aged % Aged % Aged % Above the age of % Self employed % Permanent employment % Casual employment % Non-work % Married % Single % Junior high school graduate % High school graduate/ Junior college graduate % College graduate and above % Hokkaido/Tohoku % Kanto % Chubu % Kinki % Chugoku/Shikoku % Kyusyu % Below 2 million % Between 2~3 million % Between 3~4 million % Between 4~5 million % Between 5~6 million % Between 6~8 million % Above 8 millon % Non- Work Table.4 The Difference between actual retirement age and desired retirement age (%) N Desired retirement age under ~ ~69 over 70 Under 59 years old Actual 60 years old Retirement years old age 65 years old Over 66 years old Cessation from work Disemployment All Source)Survey on Jobs and Living Conditions for Employees who passed their statutory Retirement Ages,

24 Table 5.1 The Difference between actual job status and desired job status (I) [All] (%) Desired job status Other Self Full-time Part-time activities Hope Total employed worker Worker (NPO,Volunt retirement eer) (n) Actual job status Self-employed 6.6% 42.9% 12.1% 0.4% 3.0% 1.5% Full-time worker Part-time Worker 8.6% 5.2% 49.6% 2.6% 8.5% 1.3% 20.5% 6.8% 11.8% 39.9% 31.8% 4.8% Non-working 64.3% 45.1% 26.5% 57.1% 56.7% 92.4% Total 100.0% 100.0% 100.0% 100.0% 100.0% Table 5.2 The Difference between actual job status and desired job status [by age brackets] [Age:under the age of 64] (%) Desired job status Other Self Full-time Part-time activities Hope Total employed worker Worker (NPO,Volunt retirement eer) (n) Actual job status Self-employed 5.9% 34.4% 13.6% 0.4% 3.5% 1.3% Full-time worker Part-time Worker 11.7% 8.3% 53.9% 3.0% 11.3% 0.5% 20.9% 6.3% 11.7% 33.6% 27.5% 2.7% Non-working 61.5% 51.0% 22.1% 63.0% 57.7% 95.5% Total 100.0% 100.0% 101.3% 100.0% 100.0% 100.0% [Age:over the age of 65] (%) Desired job status Other Self - Full-time Part-time activities Desiring Total employed worker Worker (NPO,Volunt retirement eer) (n) Actual job status Self-employed 6.9% 46.7% 11.9% 0.5% 2.7% 1.5% Full-time worker Part-time Worker 7.1% 3.7% 46.5% 2.4% 6.7% 1.5% 20.3% 7.1% 11.9% 43.9% 34.5% 5.4% Non-working 65.7% 42.5% 29.7% 53.2% 56.1% 91.6% Total 100.0% 100.0% 100.0% 100.0% 100.0% 24

25 Table 5.3 The Difference between actual job status and desired job status [by educational background class] [Academic background: College graduate and avobe ] (%) Actual job status Total Full-time worker Part-time Worker Other activities (NPO,Volunt eer) Desiring retirement Self-employed 6.1% 36.5% 9.1% 0.7% 2.0% 1.5% Full-time worker Part-time Worker 11.2% 6.9% 54.3% 2.7% 11.8% 1.2% 21.0% 6.9% 9.7% 42.3% 27.1% 3.1% Non-working 61.8% 49.7% 26.9% 54.3% 59.1% 94.2% Total 100.0% 100.0% 100.0% 100.0% 100.0% [Academic background: below college graduate] (%) Desired Job Status Actual job status Total Selfemployed Selfemployed Full-time worker Part-time Worker Other activities (NPO,Volunt eer) Hope retirement Self-employed 6.1% 49.3% 16.4% 0.3% 4.6% 1.3% Full-time worker Part-time Worker Desired job status 11.2% 3.6% 44.7% 2.4% 4.5% 1.1% 21.0% 6.4% 13.7% 36.7% 36.4% 4.5% Non-working 61.8% 40.7% 25.2% 60.6% 54.5% 93.1% Total 100.0% 100.0% 100.0% 100.0% 100.0% Table 6 The Difference between actual job status and desired job status (II) [All] (%) Dedired job status Actual job status Self - employed Full-time worker Part-time Worker Other activities (NPO,Volunt eer) Desiring retirement N Self-employed % 21.1% 1.4% 5.2% 9.9% Full-time worker % 69.0% 6.7% 11.6% 6.7% Part-time Worker % 8.5% 52.2% 22.3% 12.9% Non-working 2, % 4.6% 18.0% 9.6% 61.3% Total 3, % 13.0% 21.1% 11.6% 44.6% 25

26 Table.7 The relationship between job status and other various variables All Actual Job Desired Job Full-time worker Part-time worker and Retirement others Selfemployed Selfemployed Full-time worker Part-time worker and Retirement others Healthiness Academic background Business size of the former employment Experience of the former employment Regional block Non-earned income n=3020 n=189 n=201 n=624 n=2006 n=261 n=309 n=916 n=1225 Average age Healthiness (individual) Family needed caring Junior high school graduate High school graduate/junior college graduate College graduate and abobe Company with ferer than 30 workers Company with between 30 and 999 workers Company with over 1000 workers Experience as an executive Score of Job assistance programme Employment continuance from the former company Kanto Public Pension ( 10,000 yen) Non-earned income( excluding public pension)> Subsistence cost Source)Survey on Jobs and Living Conditions for Employees who passed their statutory Retirement Ages,