THE COMPETITIVE SEMICONDUCTOR MANUFACTURING HUMAN RESOURCES PROJECT:

Second Interim Report
CSM-32
Clair Brown, Editor

8. The Transferability of Case Study Findings to Other Firms
Dan Rascher and Clair Brown

8.5 Transferability in the Semiconductor Industry

The Setup

The HR Questionnaire contains questions on these specific work practice variables. As discussed in section 8.3, there are likely to be more than one set of "best practices", especially given that there are many different environments that the fabs operate in.

In light of this, we will choose 3 high performance fabs as prototypes, one each from Japan (fab J1), Asia (fab A1), and the U.S. (fab US1). A low performer from the U.S. (fab US2) and Asia (fab A2) will be chosen along with another high performer from Japan (fab J2) to be used for comparison with the prototypes.

The CSM-HR study currently contains data on 15 fabs, thus standard statistical techniques cannot be used to determine "best practices" or show that they are transferable to other fabs. The practices used by the 3 prototype fabs will be considered the "best practices" for this exercise in transferability. The practices will be compared to the 4 non-prototype fabs to help determine which practices are necessary and sufficient and which are complements and substitutes. The environmental constraints will be discussed as possible reasons for differences in performance across fabs with similar work practices.

This is analogous to the determination of the bundle X1 and then a test to see if high performing fabs (fab J2) other than the prototypes have these practices. Also a comparison to the low performers (fab A2 and fab US2) would help distinguish which practices are necessary and which are sufficient. We would expect fab J2 to have many of the same practices as the high performers, especially fab J1. We might expect fab A2 and fab US2 to be using different practices than their high performing counterparts, fab A1 and fab US1, respectively.

Key Variables

The major components of the HR system for the high performers are shown in Table 8-2 and are briefly summarized below. Core variables are practices that both high and low performing fabs use. Necessary variables are practices that are common across high performers, but not common across low performers. Sufficient variables are those practices that are fulfilled in different ways by high performers (i.e., equipment maintenance tasks can be performed by production workers, equipment engineers, or operators and technicians, and still reach high performance.).

c - core variables.

n - necessary variables.

s - sufficient variables.

* - production workers perform both operator and technician tasks.

Incentives and Monitoring

In the environment of the U.S., it appears that profit-sharing and ESOP are core practices, while individual and team performance bonuses may be considered sufficient practices for methods of compensation. In general, the high performers pay their workers more. A high wage may be a necessary condition to attract and keep the best workers and become a high performer. Highly skilled workers, and a high pay scale that increases with tenure appear to be complements as evidenced by fab US2, which has highly educated workers, but pays low wages and is a low performer. If a fab has highly skilled workers, they must be compensated accordingly or it will be more difficult to attain high performance.

There appears to be more than one set of "best promotion practices," which vary across and within regions. The determinants of promotion might be substitutable since there is more than one way to achieve high performance given the promotion practices. For engineers, output and skill level seem to be common across regions for all levels of performers.

Skill Building and Knowledge Creation

The top fabs provide SPC and cleanroom training in both the classroom and OJT, while the low performers don't. The top fabs provide training in both the classroom and OJT and more of it overall. Both the amount and type of training may be necessary conditions for high performance.

The division of labor tends to be regional. Equipment maintenance tasks are carried out by the production workers, who perform both operator and technician tasks in Japan, by both the operators and technicians in the U.S., and by the equipment engineers in Asia. SPC is done by the production workers and both types of engineers in Japan. In the U.S., some of the basics are performed by the operators, but the technicians and process engineers do the bulk of the SPC. The Asian fabs have the equipment and process engineers carry out the SPC tasks. These job assignments appear to be sufficient practices. Given that the assignment of job tasks is regional, it is possible that labor market institutions play a significant role in the determination of job tasks. Japanese fabs hire men and women with technical high school degrees. Asian fabs hire women for the operator position and men for the technician position, both with high school degrees. The labor supply is tight for these fabs and wages are low. Fabs in the U.S. hire men and women for both the operator and technician position. Operators are required to have high school degrees and technicians must have technical AA degrees.

The high performers have well-defined career ladders for both process and equipment engineers. However, this practice is not widespread across the other fabs in the sample, and so it is likely that it is a necessary practice. For technicians and operators (production workers in J1), fab J1 and fab US1 have well-developed career ladders, but fab A1 does not.

Work Organization

It appears that the high performers use job rotation, shift rotation, and teams, while the low performers do not. Thus, these practices can be considered necessary to attain high performance. The Asian fabs have the least experienced workers as operators, largely reflecting gender roles. These workers are trained in depth on a few tasks instead of on many tasks. Despite these limitations, fab A1 manages to be a high performer by complementary practices such as having a high ratio of equipment to process technicians to help the operators at each piece of equipment. The Asian division of labor seems to be substitutable, as well as sufficient, with the Japanese system in which production workers obtain substantial skill breadth. The Asian fabs also have relatively low wages and thus can use more highly skilled workers.

All of the fabs are on a 24 hour/7 days a week work schedule, except fab US2, which runs 24 hours a day for 6 days a week. Fab J1, fab J2, fab US2, and fab US1 (during the week) have 3 shifts a day for 8 hours per shift (8.75 for fab J2). Fab A1 and fab A2 are on a 2 shift schedule for 12 hours per shift. The shift scheduling seems to be regional and doesn't seem to affect the performance level of the fabs, except at fab US2, where they might have start up and shut down difficulties.

The U.S. fabs have more teams overall than any of the other regions. QIT's might be substitutable with other practices, like SDWT's or CFT's, and thus be sufficient.

8.6 Conclusion

A summary of important workplace practices shows that promotion based on output and skill level and compensation including profit sharing and ESOP's seem to be core practices. Sufficient practices include skill pay, suggestion pay, non-monetary pay (awards, gift certificates, etc.), individual and team performance bonuses, gainsharing, job task assignment, and career ladders for production workers. The division of labor is influenced by environmental variables such as cultural norms and the skill of the labor force. High pay levels, an abundance of classroom and OJT in SPC and cleanroom, along with the use of job rotation, shift rotation, teams, career ladders for engineers, and a 24 hour, 7 day schedule appear to be necessary practices.

Fab US2 faces approximately the same environmental variables as fab US1. They should be able to achieve high performance by transforming their HR system. Their compensation system needs to include individual and team performance bonuses, and the wages of a their skilled labor force should be raised to be commensurate with their level of skill. They should institute teams and cross-training for job and shift rotation along with initial training in both the classroom and on-the-job, SPC by operators, technicians, and process engineers in order to become a high performer. Since demand appears to be sufficient to warrant a 7 day schedule, a move to a 7 day work week could improve the performance of fab US2.

Fab A2 is comparable to fab A1 in terms of environmental and cultural constraints. Since fab A2 is a lower performer than fab A1, there is room for improvement. Fab A2 uses the same methods of compensation practices as fab A1, but the pay scale data, which could be a factor, is unavailable to the HR study. Tenure is used as a promotion determinant in fab A2, but not in fab A1, where skill level and output goals receive higher weighting than tenure. Cross-training and QIT's for technicians along with an increase in the ratio of classroom to OJT for engineers could lead to higher performance.

Our sample doesn't allow us to make comparisons between a high and low performing Japanese fab. However, we can note the differences between the high performing Japanese fab and the high performing U.S. and Asian fabs to gain insight into the Japanese workplace. Production/tech workers in J1 gain skill breadth and depth, which allows for the flexible use of the skilled workers to meet internal demand fluctuations. Skill development occurs through structured, on-the-job training, and the production workers' job tasks involve more complex activities and problem solving as they move up an explicit career ladder (Brown and Reich, 1996). By the time a production worker has fifteen to twenty years of experience, he (sometimes she) has the skills and job tasks of a technician. The steep experience-wage profile ensures that the more skilled experienced workers are paid more. These career ladders create an incentive for workers to increase skill depth and breadth while the broad job category provides managers with workforce flexibility.

This paper has helped to lay the foundation for developing an approach to determining transferability of case study research to other domains. There may be more than one set of "best practices" depending on the environmental variables. A test of transferability would initially involve determining the set of "best practices" given a specified environment. A comparison with high performing fabs would be made to determine whether the "best practices" are in place. Also, low performers would be tested to see that they aren't using the "best practices". Because of the low sample size in the CSM-HR study, it is difficult to perform any statistical tests to prove transferability. The exercise above was based on both quantitative and qualitative data. Although no formal tests were carried out, it helped determine some of the "best practices" and looked at which practices are necessary, sufficient, complements, and substitutes.

End of Chapter 8

Go to Chapter 9
Go References for this Chapter
Go to Table of Contents for this Chapter
Go to Table of Contents for the CSM-HR Interim Report