THE COMPETITIVE SEMICONDUCTOR MANUFACTURING HUMAN RESOURCES PROJECT:

Second Interim Report
CSM-32
Clair Brown, Editor

Chapter 3: Headcount and Turnover
Dan Rascher

3.4 Engineers

Table 3-8. Employment Levels Across Fabs

Engineering headcount is higher in the Asian fabs than in the U.S. fabs. Unlike technicians, there are more engineers in the logic fabs in Asia than in the memory fabs. However, because of the low sample size, the difference is insignificant.

Table 3-9. Coefficient of Variation 1990 - 1994

Table 3-9 shows a measure of the variability of employment over time for engineers. As with operators and technicians, employment levels fluctuate more in the Asian fabs than in the U.S. fabs. Asian engineers face the same variability in employment regardless of which type of product they help manufacture, memory or logic. Similar to operators and technicians, engineer employment fluctuated more in the Asian fabs in our sample than in the U.S. fabs, where engineer employment exhibited a steady rise. The higher turnover in the Asian fabs than in the U.S. fabs for engineers, is partly explained by higher overall employment levels and greater competition for labor. Again, the number of quits is less than or equal to the number of new hires.

Table 3-10. Annual Turnover Information (levels)

Table 3-11. Functional Headcounts - % of total

Table 3-11 shows that there is a slightly higher percentage of equipment engineers than process engineers in Asian fabs. Conversely, there is a higher percentage of process engineers than equipment engineers in U.S. logic fabs. There is a much higher ratio of equipment to process technicians than equipment to process engineers in both regions. Technicians are needed more for machine cleaning and maintenance than for work on the process. Engineers are the principal employee involved in the design of processes. A fab may need a certain number of technicians and engineers per piece of equipment, but a substantial number of additional engineers are needed for process issues.

3.5 Temporary Workers

Our sample of fabs exhibited a limited use of temporary workers. Only one of the nine fabs in this sample of U.S. and Asian fabs uses temporary operators. It is a high performing fab, and about 30% of the workers are temporary employees. Many of them have been with the fab a very long time as temporary employees and usually hold day jobs to supplement their night shift work as operators. Of the additional five fabs in our larger sample, which includes European and analog manufactures, two use temporary workers. One firm has about 33% of its operator work force as temporary, but the share is decreasing. Another firm fluctuates between 10% and 30% of its operator workforce being temporary.

One of the high performing fabs in our sample uses temporary technicians for about 30% of its technician workforce. It also has about 3% of its engineer workforce as temporary employees. One of the fabs noted that the temporary pool of workers is often used to fill in for absenteeism of engineers and technicians.

3.6 Occupation Ratios

Table 3-12 shows the distribution of the internal ratio of some job categories. The number of operators to engineers in memory fabs exceeds the same ratio in logic fabs; however, U.S. logic fabs have many more operators to engineers than Asian logic fabs. The ratio of operators per technician is roughly the same across product types, about three or four operators for every technician. The employment levels of both equipment technicians, the largest technician job specialty, and operators are likely to be highly influenced by the number of pieces of equipment. Thus, the ratios are similar across product types.

The number of technicians per engineer is greater in U.S. logic than Asia logic fabs. Asia memory fabs have over three times as many operators per supervisor as Asia logic fabs. The Asian ratio of engineers to managers is much higher than in U.S. fabs. In our sample, for a given number of operators and technicians, it appears that logic chip manufacturing requires more engineers, supervisors, and managers than memory production. In other words, the operator to technician ratio is roughly constant throughout Table 3-12. The ratio of operators and technicians to engineers (supervisors also) is lower in the logic fabs meaning that more engineers (supervisors also) are used in logic for a given number of operators and technicians. Since the engineer to manager ratio exhibits no dominant pattern across product types, and there are more engineers in logic, then there are more managers also.

As far as regional differences are concerned, the U.S. fabs use two or three times as many operators per engineer and technicians per engineer as Asian fabs. As expected, the ratio of operators to technicians doesn't vary across geographic regions. The fraction of operators per supervisor is roughly the same across regions at 13 or so for logic producers. U.S. fabs use fewer engineers per manager than their Asian counterparts. For a given number of operators, technicians, supervisors, and managers, the U.S. logic fabs use fewer engineers than the Asian logic fabs.

Are employment ratios related to the performance metrics? Even with our small sample, some interesting correlations emerge as shown in Table 3-13. A high operator to supervisor ratio is associated with more defects in logic production and lower direct labor productivity for both products. A high ratio of operators to engineers is associated with more defects, but an increased throughput, in logic fabs.

Table 3-13. Correlations of Occupational Ratios to Performance Metrics

3.7 Conclusion

In general, memory fabs have more operators but fewer engineers than logic fabs. Asian fabs have more workers of all types than U.S. fabs in our sample. There is less employment stability in the Asian fabs than in the U.S. fabs. Turnover (especially quits) is higher in the Asian fabs than in the U.S. fabs, partially reflecting the use of female operators who leave the workforce early to start families, and the local competition for workers. One example of a less stable work situation comes from one fab, which is trying to decrease its headcount by 40% while it doubles its productivity. The use of temporary workers is limited in our sample. The few fabs that do use them have about 30% of their operator and technician workforce and about 3% of their engineering workforce as temporary.

There is a higher ratio of equipment to process technicians than equipment to process engineers because of the cleaning and maintenance activities performed by technicians. On the other hand, engineers are the primary workers involved in process design, implementation, and testing.

Logic chip manufacturing uses more engineers, supervisors, and managers per operator and technician than memory chip production. In the U.S., logic producers use fewer engineers than Asian logic manufacturers per total worker. A higher number of operators per supervisor is associated with lower performance in defect density and direct labor productivity. A high level of operators per engineer is correlated with lower performance in defect density, but higher stepper throughput.

End of Chapter 3

Go to Chapter 4
Go to Table of Contents for the CSM-HR Interim Report