Division of Research September 1985 Graduate School of Business Administration The University of Michigan PROMOTION PATTERNS IN A JAPANESE TRADING COMPANYS Working Paper #440 Vladimir Pucik University of Michigan FOR DISCUSSION PURPOSES ONLY None of this material is to be quoted or reproduced without the expressed permission of the Division of Research.

I P I i

ABSTRACT This article describes promotion patterns and the resulting career timetables among managers in a large Japanese trading company. The observed patterns are contrasted with several propositions on promotions in Japanese organizations. Similarity with patterns reported in other countries is also discussed. A relatively early and irreversible "elite" identification is observed.

(

Despite often contradictory opinions about Japanese management practices, most observers agree that the key role in carrying out the activities of a typical Japanese firm is played by middle managers. For that reason, a better understanding of the factors that shape the careers of Japanese middle managers can help to learn how these organizations operate. A review of the literature reveals, however, that little empirical work examining the careers of Japanese middle managers has actually been done. It was pointed out that the main reason for the lack of empirical studies in this area is, perhaps, the difficulty of gaining access to personnel data on management-class employees (1), Even Japanese scholars have to rely on anecdotal evidence or "opinion poll" questionnaire surveys. However, such an approach does not permit a "within-cohort" analysis of promotion patterns and thus what is in many Japanese corporations the very essence of their competitive promotion dynamics remains virtually uncharted. In that sense, the cooperation offered for this study by a leading Japanese corporation made the data collected in this study unique. The study analyzes promotion patterns among middle managers in one of the largest Japanese trading firms (2). The objective is to illustrate the basic patterns of promotion timetables through a longitudinal analysis of promotion data. The empirical evidence is also compared with assumptions about the career 1

progression of Japanese managers that advanced by several descriptive studies of organizations in Japan. PROMOTIONS IN JAPANESE ORGANIZATIONS Essentially, there are only two ports of entry into a large Japanese trading firm, matched with two categories of recruits. The first port of entry is designed for new high-school or junior college graduates, mostly female, leading primarily to secretarial or other support jobs; the second is for male graduates freshly out of college, slated for managerial-level jobs. Recruitment of female university graduates is still exceptional. Nearly all promotions to management jobs are from within the firm, a system described as Internal Labor Market (3). A similar employment system applies to nearly a third of the Japanese workforce (4). All college graduates are hired under the implicit assumption that they remain with the company until they reach the compulsory retirement age, usually between the ages of 55 and CO. During most of their careers, employees are granted annual seniority pay increases, in addition to salary increases based on merit. The maintenance of so-called "life-time employment" (a somewhat incorrect but widely used term) is one of the principal - and publicly acknowdledged -- concerns and practices of the corporate management. Typically, all of the graduates who enter in the same year are promoted "automatically" during the first 12-15 years of their employment. After that, formal differences in promotion 2

patterns become visible. Naturally, some in the cohort reach the executive suite, while others have to retire from the lower ranks of the corporate ladder. What set of promotion rules, what timetables govern this process of differentiation? Marsh and Mannari have suggested that seniority and the level of education are the two most important determinants of promotions in Japanese firms (5). However, this does not say much about patterns of promotions in the most common situation, namely about promotions within a cohort composed of employees with equal tenure and of similar educational background. Nor does any other existing research provide any clues in this regard. We have learned about prewar personnel policies in a food industry firm, about the careers of junior white-collar employees in a Japanese provincial bank and in a department store, as well as those of Japanese civil servants and employees of government-owned corporations, but very little about the careers of middle-level managers in private business firms (6). Although the lack of "hard" data is one of the reasons for this study, at least two distinct hypotheses attempting to answer the question raised above have emerged from the past research. One, a minority view, proposes that even though formal selection does not play a role in differentiating a cohort until later in Japanese employees' careers, distinctions within the age group are established and maintained within a few years of recruitment and before any formal rank differences appear (7). It is argued that those who are identified as potential "stars" receive preferential assignments that may give them a decisive edge in 3

accumulating the experience record needed to earn a formal promotion later on. Implicitly, this view suggests that the promotion process in Japan is similar to processes observed in Western organizations. On the other hand, a larger group of scholars claims that a low and careful evaluation, rather than an early identification of "stars", is one of the key features of the so-called "Japanese management system" (8). This is often referred to as the "escalator system", in which young graduates of prestigious universities are virtually guaranteed promotion as they ride slowly, but surely to the top. The promotion "escalator" moves up at the same speed for all who enter the company in the same year, so there is little room for uncertainty or discomfort about career prospects. A modified version of this proposition holds that "automatic promotion ends and competition frequently begins between the rank of section head and department head" (9). That would leave plenty of time for what one writer calls a "return match", i.e., a fair chance for those who had received a lower mark to regain a top-rank evaluation (10). Obviously, this is not likely to happen under the conditions envisioned by the minority opinion. So far, neither side has presented enough evidence (for reasons stated already) to resolve the contradiction between the two positions. The study presented in this article was therefore designed to shed more light on this matter, as well as to examine other aspects of careers in a traditional Japanese business 4

organization. The investigation focuses on the careers of middle managers in a large Japanese trading company, in particular on their promotion patterns between 1970 and 1982. SAMPLE AND DATA The data were col lected in trading company T which, with a long history and thousands of employees is one of the leading trading firms in Japan. The bulk of the data was collected in 1980 and updated in 1983. On both occassions, additional interviews were conducted with a cross-section of managers in the firm. All figures presented here were computed from various personnel records provided by the company officers at its headquarters in Tokyo. For reasons of confidentiality, the absolute values of data are not reported and some other indexes were modified in a fashion that does not influence the findings. About two-thirds of all employees in the firm are males hired immediately after they graduated from Japan's prestigious universities, the rest are females assigned to secretarial jobs, security, maintenance and related workers. The formal hierarchical system in the company is fairly complex, the critical ranking that determines compensation is that of "status" - a form of grade not related directly to any specific job assignment. A similar system ranks job titles. Generally, each status grade is linked with two or three classes of job titles. A convenient analogy to such a structure would be a typical military organization. Promotion to a given status grade opens 5

eligibility for a range of job classes, but the match between status grades and job titles is not perfect. An assignment "outside" the grade is not common, but it does happen. All male university graduates (called "regular employees" in the text below) are initially ranked according to the year of entry into the firm. Their positions within the cohort are formally undifferentiated until their mid-thirties -- the time for promotion to the "status" of Manager. Above the grade of Manager, there are three other non-executive middle management grades, and each person has to spend at least four years in each of the grades before moving up to the next. The status grades are ranked as follows: GRADE STATTIS TITLE 1 General Manager 2 Deputy General Manager 3 Assistant General Manager 4 Manager 5 Assistant Manager 6-8 Ordinary FTplouyee A manager with direct supervisory responsibility is called a "line" manager, the others are either "professional" or "staff" managers. The professional managers (e.g. foreign exchange traders, insurance brokers, or system analysts) occupy nonsupervisory positions that require specialized expertise, while the staff managers cover all other non-supervisory assignments (11). Within a grade, salary differentials between line and staff from the same cohort are limited to a nominal supervisory supplement. 6

In 1970, the employees who belonged to any of the four upper grades accounted for 36.3 percent of all regular employees. This proportion reached 49.4 percent by 1982, primarily due to the cut in recruitment during the recessions following the two oil crises of 1973 and 1980. THE STRUCTURE OF PROMOTION OPPORTTUNITIES The basic structure of middle management positions in Company T is exhibited in TABLE 1. Data are shown only for domestic offices and branches that employed just over 70 percent of all managers in the firm. According to other company data, the grade distribution of managers in overseas branches and affiliates is similar to that in the parent company in Japan, but staff/line distinctions do not apply. Nearly all expatriates of management rank are assigned, at least nominally, to line management positions. TABLE 1: STRUCTJURE OF MANAGERIAL POSITIONS, 1982* RANK GRADE 1 GRADE 2 GRADE 3 GRADE 4 Managers/ 5.6 10.5 30.1 53.8 All Managers Line Managers/ 9.6 14.7 36.8 38.9 All Line Managers Staff Managers/ 1.8 6.5 22.7 69.5 All Staff Managers Line Managers/ 84.6 70.9 62.8 37.2 All Managers Staff Managers/ 15.4 29.1 37.2 62.8 All Managers * Domestic operations only 7

The resulting distribution of grades does not seem to he particularly unique. Over one half of all middle managers are ranked in grade 4 which is the lowest management rank. Only 5 percent of managers are classified in grade 1 as general managers. Clearly, the hierarchy is steep and the "escalator" anology does not seem to be supported. A similar kind of distribution data was used by Marsh and Mannari to estimate promotion chances in several Japanese manufacturing firms. "Other things being equal, the larger the number of positions at a given level, the greater the objective chance for those in the next level" (12). Following this logic, one could say that about 11 percent of employees in grade 4 have an "objective" chance of being promoted up to the rank of General Manager. Yet, it will be shown later that the actual promotion ratio during this period was close to 25 percent. The reason for the discrepancy is that "other things" are rarely equal. Without taking into account the growth of the organization, the number of people in each cohort, the promotion eligibility rules for each grade, the retirement patterns (vacancies), etc., the statistics based on the "single-time stock figures" can be misleading. To get a better understanding of the trends shaping careers, more detailed data have to be examined. One possibility would be to follow the proposition concerning the differences in job assignments, but the data on the individual job histories of managers in Company T were 8

unfortunately not made available. However, an aggregate set of data shown in TABLE 1 indicates differences in promotion opportunities depending on whether a manager is employed in a line or a staff/professional position. According to the information collected in interviews it is infrequent for a manager to receive a higher-level line assignment without a corresponding line experience on the lower level. Also, given that in Company T the customary tenure in one job is four years, the same as the minimum stay in each of the higher grades, it does not seem probable that a typical "fasttrack" employee can accumulate both line and staff/professional experience during his stay at one grade level. As a result, managers in company T do not often cross the "line/staff" boundaries, and those having a record of line managerial responsibilities are 5 times more likely (everything else being equal) to reach grade 1 than those assigned to staff and professional positions. The shortage of upper-level staff jobs suggests that, without line experience, promotion chances decline considerably. In fact, the difference in the number of status positions between grade 4 and grade 3 is nearly all at the expense of the staff personnel. The data indicate that an assignment into a. staff position may have a long-term negative impact on an employee's career. This situation is not peculiar to Japan; it has been observed in American organizations as well (13). What is noteworthy, is that on a short-term basis, e.g. within a grade, such differences are 9

hidden by equality in compensation. It seems, however, that most employees are very much avere of what the long-term rules are, even though officially the data on line/staff distribution are "top secret". Recently, company T attempted to formalize the professional career track and target candidates for future assignments as professional managers early in their careers. One of the key purposes was to improve the quality of professionals by exposing them to a more focused on-the-job training. However, resistance to early formal track differentiation from the employees forced the company to shelve the idea at least for the time being. Naturally, employees did not opposed the improved training. They were concerned that they will end up being locked out of supervisory positions. The data suggest that their fears may be justified. COHORTS AND CAREEmRS It was suggested earlier that in order to estimate correctly the promotion opportunities, the size of the cohorts coming up for promotions has to be considered. FIGURE 1 shows the relative sizes of the age cohorts in 1973 and 1979 for employees in the upper grades (14). Observing the rigidity of the promotion system in Comnany T, it would seem that either the promotion odds will change inversely with the size of the cohort, or the number of the available positions must be adjusted accordingly, so the fluctuations in cohort sizes can be accomodated. 10

INSERT FIGURE 1 HERE In the last twelve years the total number of employees in the firm increased by 10 percent but the number of employees awarded a status higher than grade 4 increased by 46 percent. This is in direct proportion to the increase of older employees in the firm; this suggests that the number of upper-level positions was indeed increased. Similar strategies for coping with an aging workforce were observed in other Japanese firms (15). Nevertheless, a more specific analysis of the data reveals substantial differences in how this adjustment was actually made at each of the higher grades. TABLE 2 presents promotion rates since 1970 for each grade as wel 1 as the degree of their yearly variations between 1970 and 1979. TARLE 2: MANAGEMENT PROMOTION RATES, 1970-1982 RANK INCREASE IN NUMBER AVERAGE PROMTION VARIATION OF POSITIONS RATE RATIO* GRADE 1 151% 12.8% 0.45 GRADE 2 126% 9.8% 0.36 GRADE 3 146% 13.1% 0.27 GRADE 4 153% 14.7% 0.19 * Variation Ratio = Standard Deviation/Average Promotion Rate The promotion rate at grade 4 is the most stable, but the variation ratio gradually increases as one moves up the 11

hierarchy. This indicates that in grade 4 the more candidates there were, the more slots were made available, while in grade 1 the promotion odds varied rather dramatically. The number of slots were held relatively stable, while the size of the candidates pool 's pool fluctuated with the size of the cohort. The data presented in TABLE 2 also illustrate the screening function of the Personnel Division. Until grade 3, nominations from division managers are al 1 that is needed to promote those who have already fulfilled the required tenure in the grade. However, the promotions to grade 2 have to be approved by Personnel and, as a result, the promotion rates are cut by more than 25 percent. However, advancing to grade 1, the odds for success increased again for those who passed the previous rigorous check. Given the emphasis on seniority in Japanese organizations commented on by many writers (16), one may believe that when positions are limited, the more senior person will get the promotion. For Company T, however, the data suggest otherwise. Similarly to the bank observed by Rohlen (17), some outstanding performers in the cohort are promoted at the earliest possible date, the bulk during the next two years, but the rest never catches up. The resulting distribution of status in the cohort is summarized in FIGURE 2 (the average is 1973-1982). INSERT FIGURE 2 HERE 12

The flatter the downward scope of the curve, the fewer people are promoted in a given time period. The rapid decrease in probability of promotion is clearly visible. The average promotion rates indicated in TABLE 2 thus underestimate the chances of those advancing rapidly and overestimate the chances of those lagging behind. THE CAREER TREE The preceding two sections discussed the basic structure of promotion opportunities in Company T as it applies to cohorts or classes of employees. However, in order to understand the impact of this structure on individual careers, the focus of the analysis must shift to the individual level. For that purpose, all promotions into and within the four top grades between 1973 -1979 are combined into a model career tree in FIGURE 3. The career tree indicates the probabilities (p) of grade promotion given a manager's years of service in the company (numbers in the boxes) and his current ranking. In other words, the probabilities indicate the odds that employees who, after they reached a specific grade at a specific time, wi 11 move to a higher grade along a particular transition line. INSERT FIGURE 3 HERE Although the personnel systems in Japan seem to have many facets that make them appear very different from those in the U.S., the career tree in Company T resembles the "tournament 13

career model" described by Rosenbaum in the case of an American firm (18). Roth models allow "recovery" of a fallen ranking only in exceptional cases. Managers in Company T are promoted according to the standard four year timetable or they fal even further behind in the "tournament". Gradually, most managers are left behind. To become a top grade 1 manager, an employee has to secure nomination to grade 4 at the earliest time possible. During the period investigated, on average, six percent of the cohort members achieved a "fast-track" promotion, while an additional seven percent reached grade 1 a year later. However, those who reached grade 4 one year after the "fast-track" group had only a five percent chance to get to grade 1 by their 27th years of service, and thus to remain in the running for an executive position. During the six years examined, only one employee was promoted above grade 1 into an executive position after being delayed for more than a year at any middle management level. Usually, the "late arrivals" to grade 1 are dispatched to a management post in an affiliated firm. Three basic career timetables emerge from patterns presented in FIGURE 3. The first timetable applies to the group of elite managers with mostly line experience. All of them reach grade 1 by their 27th year at the latest. Some managers in the second group which is composed of the remaining line managers and some professional managers also reach grade 1, but most retire from grade 2. The remaining group composed mostly of staff managers retires in the two lowest management grades. 14

Clearly, the fifteenth years of service is the "make-orbreak" time for managers in Company T. More than half of the cohort members who were not promoted to grade 4 by the fifteenth year lost their chances of ever getting to grade 1. One more year of delay eliminates the chances for grade 2 as well. On average, in addition to nearly 7 percent of employees who never become Managers, 20.2 percent exited from the firm at grade 4, 21.7 percent at grade 3, 27.5 percent at grade 2. Altogether about 24 percent of the cohort was promoted to grade 1, which is more than twice the estimate based on the structure of positions in the company discussed earlier. About a fifth of those who reach grade 1 are eventually promoted to executive jobs. It is interesting to note that even though promotion to grade 2 is "objectively" the most difficult, the largest group of employees retires from that grade. The promotion rate might be the lowest at grade 2, but eligibility for promotion is the longest. Even after 5 years' delay at grade 3 there is still a 44 percent chance of being promoted to grade 2, but a similar delay at grade 4 indicates the end of any meaningful career opportunities. When the status differences among the cohort members become visible after fourteen years of service in the firm, for a vast majority of the employees in Company T, it is already too late to get back on the "fast-track" by trying harder. There is indeed an intense competition among Japanese managers in their thirties as suggested by Clark (19), but in Company T, this competition is of two kinds. One among the elite competing for 15

future promotions, the other among the rest of the cohort who fear slipping too fast too soon. The patterns observed in the career tree do not support the "long-term evaluation" hypothesis. Seeing the sharp and "final" differentiation in grade 4, it seems highly probable that early informal recognition through preferential job assignments occurs - with few "return matches" scheduled for later on. CONCLUSION The data indicate that to decribe the promotion process in Japanese firms as a comfortable "escalator" may be highly misleading, at least in the case of Company T. A crowded elevator is perhaps a more appropriate analogy. After rising through the lower floors of the corporate pyramid, the career elevator stops frequently. The door opens and there is a lot of jostling. Those who are unlucky enough to be close to the door, or those with less sophisticated elbow techniques are pushed out and have to find a place in other, slower elevators. As is often the case in Japan, even this goes on with a polite smile and bow towards those showed out the door. It is not appropriate to generalize findings from a single case to other Japanese organizations. On the other hand, given that Company T is a well-known trading firm (a type of firm often considered unique to Japan), the fact that the careers observed in Company T were on a number of counts (e.g., early and fixed status differentiation, unequal treatment of staff and profesional employees, emphasis on merit over seniority, etc.) 16

different from what has been presented as the "typical" career of Japanese managers, encourages some scepticism about the stereotyped images of management in Japan, as well as underscores the need for deeper and wider cross-cultural studies of management career progression. 17

REFERENCES 1. H. Hazama. "On the International Controversy About the Characteristics of Japanese Management: Comments" (in Japanese), Organizational Science, Vol. 3, No. 3, October 1979, pp. 67-69. 2. For a detailed description of Japanese trading firms see Y. Tsurumi, Sogoshosha:Fngines of Export-based Growth. Montreal: Institute for Research on Public Policy, 1980. 3. R.E. Cole, Work, Mobility and Participation. Berkeley, Cal: The University of California Press, 1979. 4. Estimates of "life-time" employment vary from 20 to 40 percent of Japanese workforce depending on the methodology used. 5. R. M. Marsh, and FT. Mannari, Modernization and the Japanese Factory. Princeton, N.J.: Princeton University Press, 1976. 6. See, for example, M. Fruin, "The Japanese Company Controversy", The Journal of Japanese Studies, Vol. 4, No. 2, Summer 1978, pp. 267-300; T. Rohlen, For Harmony and Strength. Berkeley, Cal: University of California Press, 1974; K.A. Skinner, "Aborted Careers in a Public Corporation". In D. 1W. Plath, (Ed.), Work and Life Course in Japan, Albany, N.Y: SUNY Press, 1983; M. Wakabayashi, et al., "Management Career Development: Japanese Style," International Journal of Intercultural Studies, Vol. 4, 1980, pp. 381-419. 7. Rohlen, op.cit. 8. W.G. Ouchi, Theory Z. Reading, Mass: Addison-Wesley, 1981. 9. R.C. Clark, The Japanese Company (New Haven, Connecticut: The Yale University Press, 1979), p. 116. 10. K. Koike, The Company-Wide Training: The Special Characteristics of Japanese Labor. (in Japanese), Tokyo: Nihon Keizai Kenkyu Centa, 1980, mimeographed paper. 11. In the rest of the article,, the term "staff" include professional managers unless specified othervise. 12. R.M. Marsh and H. Mannari. "Technological Implications Theory: A Japanese Test" Organizational Studies, Vol. 1, 1980, No. 2, p. 180. 13. F.H. CGoldner and R.R. Ritti. "Professionalization as Career Immnobility", The American Journal of Sociology, Vol. 75, 1969, pp. 489-502. 18

14. The time period was selected so the fluctuations in career sized can be well illustrated. 15. K. Haitani, "Changing Characteristics of Japanese Employment System", Asian Survey. Vol 18, 1978, pp. 1029-1045. 16. e.g. Marsh and Mannari (1976). 17. Rohlen, op.cit. 18. J.E. Rosenbaum, "Tournament Mobility: Career Patterns in a Corporation," Administrative Science Quarterly, Vol. 24, 1979, 220-241. 19. Clark, op.cit. 19

FIGURE 1 VARIAI1ONS IN COHORT SIZES 1979 15 101 0 01) N Co Average Size of Cohort 1973 14 18 22 26 30 Tenure (in years)

FIGURE 2 STATUS DISTRIBUTION IN A COHORT (average 1973 - 82) 1 10 14 18 22 26 30 Tenure (in years) "- ~c --- —-ce

FIGURE 3 CAREER TREE Grade 5 Grade4 Grade3 Grade 2 Grade 1 6.1 ~ % of Cohort Reaching Grade 1 at the Time 6.8 4.1 2.6 4.3 Retiring in the Grade.x --. ---. < p <.2.2< p <.5 -------.5< p < 1 p<.1 excluded % of Cohort Retiring in 20.2 21.7 the Grade * Including promotions to the executive level 27.5 23.8*

FIGURE 2 STATUS DISTRIBUTION IN A COHORT (average 1973- 82) 101 75 50 25 10 14 18 22 26 30 Tenure (in years)