CHAPTER 3................................................................................................................ 18
3.3 Research methodology
3.3.2 Cluster analysis
Cluster analysis is a statistical technique that sorts observations into groups, which have similar characteristics. Without prior knowledge, cluster analysis was applied to develop our taxonomy of strategic groups. Existing clustering algorithms can be mainly classified into two categories: hierarchical and non-hierarchical methods. In our study, we use K-means algorithm, one of the non-hierarchical methods, due to no prior knowledge with the electronic industry. Hence, cluster analysis grouping organizations by minimizing the multivariate distance between firms within group while maximizing the distance between groups, using all observed relationships among configuration-defining variables to assign firms to clusters (Hair, Anderson, Tatham, &
Black, 1992).
Besides, we wonder what numbers of clusters should be chosen. The Cubic Clustering Criterion (CCC) and Pseudo F statistic (PSF) are used to estimate the number of clusters. The Cubic Clustering Criterion (CCC) was developed by SAS as a comparative measure of the deviation of the clusters from the distribution expected if data points were obtained from a uniform distribution (Sarle, 1983). The pseudo-F statistic is intended to capture the 'tightness' of clusters, and is in essence a ratio of the mean sum of squares between groups to the mean sum of squares within group (Lattin et
al., 2003: 291). Values of the Cubic Clustering Criterion greater than 2 or 3 indicate good clusters; values between 0 and 2 indicate potential clusters, and negative values could indicate outliers. The local peak of the Cubic Clustering Criterion (CCC) should be chosen. On the other hand, relatively large values of Pseudo F statistic (PSF) indicate a stopping point.
CHAPTER 4
RESULTS AND DATA ANALYSIS
4.1 Principle components analysis
A principle component analysis was conducted to identify source configurations correlated with the financial performance indicators in the 290 firms. We apply a varimax rotation and identify six factors (eigenvalue>1), which account for 71.83% of the total variance. Figure 4-1 shows the scree plot and Table 4-1 shows these six source configurations and their associated financial indicator loadings. The significant loadings (0.50 and above) would highlighted in bold.
Figure 4-1 Scree plot
Source: this study
Table 4-1.Principle component analysis Financial Source Configuration Indicators Factor1 Factor2 Factor3 Factor4 Factor5 Factor6 Scale Knowledge Light-assets Relationship Continuity Fixed-assets
Employees 0.90 -0.07 0.14 0.03 -0.06 -0.06 Total assets 0.89 -0.10 0.18 0.05 0.05 0.02 SG&A/Sales -0.09 0.85 0.02 0.02 0.00 -0.20 R&D/Sales 0.03 0.79 -0.16 0.02 -0.35 -0.05 CGS/Sales 0.10 -0.73 -0.22 0.00 -0.05 0.00 Inventory turnover -0.13 0.03 0.86 -0.02 -0.04 -0.03 Intangible assets 0.38 0.05 0.80 0.06 -0.04 0.03 Goodwill 0.42 0.03 0.73 0.01 -0.02 -0.05 Accounts payable turnover -0.04 -0.05 -0.03 0.89 0.02 -0.11 Accounts receivable turnover 0.14 0.12 0.05 0.81 -0.18 0.28 Company category -0.15 -0.03 0.07 -0.03 0.78 0.20 Operating years 0.08 -0.09 -0.16 -0.10 0.73 -0.17 Fixed assets turnover -0.06 -0.24 -0.01 0.06 -0.05 0.75 Dep/Sales 0.49 0.04 0.04 -0.14 -0.24 -0.50 Eigenvalue 3.02 2.32 1.74 1.52 1.11 1.06 Accumulated variance (%) 0.20 0.36 0.47 0.57 0.65 0.72 Source: this study
In Factor 1, the significant indicators are related to Scale management. This includes Employees and Total assets.
In Factor 2, the significant indicators are related to Knowledge management. This includes R&D/sales, SG&A/sales, and cost of good sold/sales. From R&D/sales and SG&A/sales ratios, we could assess the firms’ efficiency of resource deployment. There is also a negative correlation between CGS/sales and Factor 2 (-0.73), indicating that good knowledge management can pay off respect to a lower cost of good.
In Factor 3, the significant indicators are related to Light-assets management, which includes inventory turnover, goodwill, and intangible assets.
In Factor 4, the significant indicators are related to Relationship management. This includes customer relationship management (accounts receivable turnover) and supplier relationship management (accounts payable turnover).
In Factor 5, the significant indicators are related to Continuity management, which includes company category and operating years.
In Factor 6, the significant indicators are related to Fixed-assets management, which includes fixed assets turnover and Depreciation/sales.
4.2 Cluster analysis
In this section, we conduct K-means cluster analysis according to six key factors in section 4.1. K-means cluster analysis is a multivariate statistical technique, which involves grouping similar objects into mutually exclusive clusters. By the K-means clustering method, each company would be classified into single cluster so that each cluster wouldn’t be overlapped. After the grouping process, each company would be accurately positioned in the cluster, which possesses the most similarities within the clusters.
According to Cubic Clustering Criterion, the best appropriate number of cluster would be four, which presents Pseudo F Statistic 44.05 and Cubic Clustering Criterion 0.11. Table4-2 shows the Cubic Clustering Criterion value.
Table 4-2 Cubic Clustering Criterion
MAXC=2 MAXC=3 MAXC=4 MAXC=5
Pseudo F Statistic 32.85 41.39 44.05 40.82
R-Squared 0.13 0.23 0.32 0.38
Cubic Clustering Criterion -4.44 -1.07 0.11 -2.49
Source: this study
From the Table 4-3, we could know six management factors of each cluster according to cluster means, which had been standardized earlier. Each cluster has a similar organizational configuration when we decomposed the management capabilities
of each company. Companies in the same cluster represent as their strength in certain managements as their weakness in certain managements.
Cluster 1 reveals their strength in Knowledge management and Relationship management rather than weakness in Scale management and Light-assets management.
Thus, we name Cluster 1 “Knights”.
Cluster 2 has no superior management capability than other clusters. In addition, they have the worst performance in the Knowledge management, Continuity management, and Fix-assets management. Thus, we name Cluster 2 “Paupers”.
Contrast to Cluster 2, Cluster 3 has superior management capabilities in the Continuity management, and Fix-assets management. However, Cluster 3 was the worst one in the Relationship management. Thus, we name Cluster 3 “Laborers”.
Unlike those Clusters above, Cluster 4 having two strength managements, Scale management and Light-assets management, but having no weakest management capabilities. Thus, we name Cluster 4 “Kings”.
Cluster4
Kings Cluster3
Laborers Cluster2
Paupers Cluster1
Knights Factors
Clusters
3.99021 -0.14822 -0.04352 -0.26419 Factor1
Scale
management
0.12082 -0.21581 -0.41365 1.80838 Factor2
Knowledge
management
2.94842 -0.08284 -0.08825 -0.15276 Factor3
Light-assets
management
0.17546 -0.11910 -0.08563 0.61441 Factor4
Relationship
management
0.01318 0.67150 -0.87757 -0.27726 Factor5
Continuity
management
-0.12089 0.12635 -0.17403 -0.00134 Factor6
Fixed-assets
management Table 4-3 Cluster Means
Source: this study
The electronic industry was categorized into four organizational configurations by K-means cluster method. Next, we use discriminate analysis to check the accuracy of the classification result. Cross-validation is done by recalculating the discriminant function for all companies other than the validated companies. Table 4-4 shows the overall hit ratio is 95.17%, which reveal that the classification of the K-means cluster is considerably fit.
Table 4-4 Classification Results used for Cross-Validation Clusters Cluster1 Cluster2 Cluster3 Cluster4 Total
Cluster1 36 0 3 0 39
Cluster2 0 98 0 0 98
Cluster3 0 11 133 0 144
Cluster4 0 0 0 9 9 95.17% ((36+98+133+9)/290) of the cross-validated firms remain correctly classified.
Table 4-5 shows four clusters of the 290 electronic companies. There are 39 firms in cluster1, 98 firms in cluster2, 144 firms in cluster3, and 9 firms in cluster4. Cluster4, comprising only 9 firms, includes many well-known firms such as Hon Hai Precision Ind Co., Taiwan Semiconductor Mfg. Co., United Microelectronics Corp., AU Optronics Corp. etc. In addition, three main telecommunication firms, Chunghwa Telecom Co., Taiwan Mobile Co., and Far EasTone Telecommunications Co., also classified into the same cluster. Next section, we would examine the performance among four clusters.
Table 4-5 Companies in the clusters1
Table 4-6 Companies in the clusters2
Leadtek Research Inc.
Pan Jit International Inc.
Amtran Technology Co., Ltd.
Infodisc Technology Co., Ltd.
Turbocomm Tech. Inc.
Prime Optical Fiber Corp.
MiTac Technology Corp.
Elite Semiconductor Memory Technology In
Precision Silicon Corp.
Asia Vital Components Co., Ltd.
ICP Electronics Inc.
Wistron NeWeb Corp.
Richtek Technology Corp.
Arima Optoelectronics Corp.
Lite-onit Corp.
Sitronix Technology Corp.
Nan Ya Printed Circuit Board Corp.
Compal Communications, Inc.
Arima Communications Corp.
Giantplus Technology Co., Ltd.
Walton Advanced Engineering Inc.
Darfon Electronics Corp.
Creative Sensor Inc.
Associated Industries China, Inc.
Table 4-7 Companies in the clusters3
Synnex Technology International Corp.
Universal Scientific Industrial Co., Ltd
Yosun Industrial Corp.
Table 4-8 Companies in the clusters4
In this section, we would examine the different performance between clusters. We use three performance criterions, which include ROIC, ROE, and Net profit to test the difference. Table 4-6, we present three average financial performance of each cluster.
Cluster4 dominate over the other clusters, which hold ROIC 13.759%, ROE 8.003%, and Net profit 15.006%. Cluster3 win the second position in three performance criterions, which were separately 12.414%, 8.003%, and 6.567%. Cluster2 have the worst ROIC and ROE, which are -20.078% and 2.749%, but have more superior performance at Net Profit than Cluster1. Cluster2 are the worst configurations according to their performance.
Table 4-9 Average performance of each cluster
According to Table 4-6, we know that Cluster4 perform the best. From Table4-7, we conduct multiple analysis of variance (MANOVA) and analysis of variance (ANOVA) tests to examine the significance of configuration-performance relationship. A multivariate test (Wilks’ Lambda=0.9455; p<0.10) indicated that the configuration-performance relationship is slightly significant, which means various organizational configurations possess different financial performances. To summarize, different configurations possess different management capabilities, which result in various financial performances. Cluster4, especially, would be the best ideal configuration in electronic industry.
Table 4-10 Multivariate Regression Analysis ROIC ROE Profit Cluster1 4.77** 1.91 161.42***
Cluster2 0.80 5.88*** 9.10***
Cluster3 4.49*** 2.69* 27.04***
Cluster4 19.21* 3.92 13.11*
F-Value 2.44 2.35 2.05 R-square 0.0249 0.0241 0.0210 MANOVA Wilks’ Lambda=0.9455 F=1.79
p≦0.10; *p≦0.05; **p≦0.01; ***p≦0.001.
Source: this study
﹢ ﹢ ﹢
﹢
﹢
CHAPTER 5
DISCUSSION AND CONCLUSION
5.1 Research Finding
Following the entire research process, this study derives some meaningful findings based on the empirical data analysis for the electronic industry. The main research findings are generalized into four conclusions below:
1. Derive the six management capabilities. According to Principle Component Analysis (PCA), we could derive six major management capabilities from the fourteen financial indicators. We decompose the six major management capabilities in Table 5-1.
2. Classify four strategy groups. Table 5-2 shows financial values of each cluster.
Cluster1-Kinghts
Strengths: Knowledge management, Relationship management Weaknesses: Scale management, Light-assets management, Continuity
management, and Fixed-assets management
There are 39 companies belong to Cluster1, whose scale are the smallest.
Most of these companies belong to semiconductor category. These companies have the least assets and employees. Besides, they get the least goodwill and the lowest inventory turnover, which would result from inefficiency of inventory management system and result in additional cost.
On the other hand, Cluster1 are good at knowledge management. They spend more capital in administration and research & development, and thus decrease the cost of good sold, which only 0.579. Besides, they have a good relationship with their consumers, which result in a high accounts receivable turnover. Cluster1 focus their strategy on Knowledge management and Relationship management.
Cluster2-Paupers
Strengths: No superior management
Weaknesses: Scale management, Knowledge management, Light-assets management, Relationship management, Continuity management, and Fixed-assets management.
There are 98 companies belong to Cluster2. Most of them deal with photoelectrics. Cluster2 have no core strategy and get the highest cost of good sold 0.851, which indicates the inefficiency of their overall process.
Cluster3-Laborers
Strengths: Continuity management, and Fix-assets management
Weaknesses: Scale management, Knowledge management, Light-assets management, and Relationship management
There are 144 companies belong to Cluster3, which are the most common form in electric industry. These companies deal with Electric component and have the highest operating years, which indicates their experience and speciality in this category. Besides, they get the highest fixed assets turnover and lowest depreciation/sales, which indicates their strength in utilizing their fixed assets to create profit. However, they have poor relationship with both their suppliers and consumers, which cause higher account payable turnover and lower account receivable turnover. In a short, Cluster3 focus on their speciality and adopt a Fix-assets management strategy.
Cluster4-Kings
Strengths: Scale management, Knowledge management, Light-assets management, Relationship management, Continuity management
Weaknesses: Fixed-assets management
There are 9 companies belong to Cluster4. Three of them are engaged in semiconductor and the other three companies deal with Internet communication.
These companies feature for their large scale and great deal of goodwill such as trademark and intangible assets such as patents, which elaborate a synergy in their overall operation. In addition, they get along well with both their suppliers and consumers, which cause the lowest accounts payable turnover and highest accounts receivable turnover. In short, Cluster4 focus their strategies on scale, light-assets, and relationship management.
Table 5-2 Average financial values of each cluster Cluster1 Cluster2 Cluster3 Cluster4
Knights Paupers Laborers Kings Employees 432.74 1417.43 866.48 13437.67 Total assets 7223862 26229058 19017172 388944216 SG&A/Sales 0.316 0.114 0.113 0.132 R&D/Sales 0.113 0.035 0.023 0.028 CGS/Sales 0.579 0.851 0.801 0.722 Inventory turnover 4.564 8.381 7.082 16.316 Intangible assets 129605 202427 70734 4226394 Goodwill 22950 71364 42522 4148644 Accounts payable turnover 72.566. 64.438 73.836 42.218 Accounts receivable turnover 8.234 5.983 4.860 8.337 Company category (Mode) 1 3 5 1 & 4 Operating years 17.4 15.2 26.2 18.1 Fixed assets turnover 5.657 9.745 17.715 2.167
Depreciation/Sales 0.040 0.077 0.037 0.197 Source: this study
5.2 Research contribution
In this paper, we start from the financial data and develop a framework to examine variously listed companies in electric industry. Based on our research framework, we use relative financial indicators to distinguish major management capabilities, which provide a new principle to look into each company. In addition, we construct four strategies groups according to six management capabilities. The main contributions of this study are discussed below:
1. Distinguish the measurement factors of management capabilities. In this study, we adopt fourteen financial indicators to conduct Principle Component Analysis (PCA), and generate six principle measurement factors to illustrate the various management capabilities. These measurement factors are Scale management, Knowledge management, Light-assets management, Relationship management, Continuity management, and Fixed-assets management.
2. Classify strategy groups within the electronic industry. Based on the six principle measurement factors, we conduct K-means cluster analysis to classify the electric industry into four type configurations. Cluster4 are the best configurations, whose strategies are focused on Scale management and Light-assets management. Cluster3 are the second configurations, whose strategies are focused on Continuity management, and Fix-assets management. Cluster1 are the third configurations, whose strategies are focused on Knowledge management and Relationship management. Cluster2 are the worst configurations, which have no superior management capability.
3. Verify the significance relationship between organizational configurations and operating performances. Based on the empirical data examination, the significant results from MANOVA test (Wilks’ Lambda=0.9455; p<0.10) proved that organizational configurations do affect the operating performances. Hence, different configurations determined by each firm would become a strategy to enhance management capabilities and obtain outstanding financial performances.
4. Derive the most profitable organizational configuration. After measuring the operating performance indicator such as ROIC, ROE, we find out Cluster4 dominate the
other clusters. Cluster4 is the most profitable organizational configuration in the electronic industry. Cluster4 get superior capabilities in Organizational-scale management and Light-assets management. Cluster4 are composed of large-scale companies such as Hon Hai Precision Ind Co and Taiwan Semiconductor Mfg. Co. These companies possess a good deal of assets and employees. Besides, they get goodwill and intangible assets such as patents and trademark which could lead to a leverage effect and generate a synergy. In addition, their inventory turnovers are more than the other clusters, which means these companies are good at selling its inventory and generating income.
5.3 Research limitation
In this paper we adopt listed electric companies as our research sample. Hence, we only acquire empirical results to support our prior hypothesis in electric industry rather than obtain a general truth, especially empirical study would vary by various industries.
In addition, we adopt the deductive approach to select critical variables provided by the TEJ Data Bank. Thus, we might ignore some meaningful financial indicators, which TEJ Data Bank did not provide and have crucial influences to our research result.
Last, this paper focused on Taiwan electric companies without considering International electric companies. Hence, we could not compare both to derive the ideal strategic group, which decrease the research contributions for Taiwan electric companies.
5.4 Research recommendation
There are two main research recommendations for future study below:
1. In this paper, we derive the ideal configuration in electric industry, which got the highest performance. However, we are wondering if there exists a penalty
relationship for configurations deviating from the ideal configuration?
2. In this paper, we distinguish six major management factors. However, we are wondering if there exists a trade-off relationship among these factors or if there exists a dominant management factor contributes to performance most.
Reference
Burns, T., & Stalker, G. M. 1961. The management of innovation. London: Tavistock.
Chandler, A., 1962. Strategy and Structure. MIT Press, Cambridge, MA.
Charles T. Horngren, Walter T. Harrision Jr., & Linda Smith Bamber, 2002. Accounting (5th edition). Prentice Hall International, Inc..
Cool, K. O., & Schendel, D. 1987. Strategic group formation and performance: The case of the U.S. pharmaceutical industry, 1963-1982. Management Science, 33: 1102-1124.
Christopher Gresov & Robert Drazin, 1997. Equifinality: Functional equivalence in organization design. Academy of Management Review, vol. 22, No. 2, 403-428.
Edelbrock, C. 1979. Comparing the accuracy of hierarchical clustering algorithms: The problem of classifying everybody. Multivariate Behavioral Research, 14, 367-384.
Fiegenbaum, A., & Thomas, H. 1990. Strategic groups and performance: The U.S.
insurance industry, 1970-1984. Strategic Management Journal, 11: 197-215.
Fisher C. 1999. Driving financial performance through the du Pont identity: A strategic use of financial analysis and planning. Financial Practice & Education, vol. 9, No. 1, 34-45.
Galunic, D. C., & Eisenhardt, K. M. 1994. Renewing the strategy-structure-performance paradigm. Research in organizational behaviour, vol. 16: 215-255.
Giddens, S. 1979. Central problems in social theory. Berkeley: University of California Press. Grant, R. 1991. The resource-based theory of competitive advantage: implications for strategy formulation. California Management Review, vol. 33, No. 3, 114-135.
Hair, J. F., R. E. Anderson, R. L. Tatham, & W. C. Black, 1992. Multivariate Data Analysis (3rd ed.). Macmillan, New York.
Hrebeniak, L., & Joyce, W. 1985. Organizational adaptation: Strategic choice and environmental determinism. Administrative Science Quarterly, 30: 336-349.
Katz, D., & Kahn, R. L. 1978. The social psychology of organizations. New York: Wiley.
Ketchen, D. J., J. B. Thomas, C. C. Snow. 1993. Organizational configurations and performance: A comparison of theoretical approaches. Academy of Management Journal, vol. 36, 1278-1313.
Ketchen, D. J., C. L. Shook. 1996. The application of cluster analysis in strategic management research: An analysis and critique. Strategic Management Journal, 17 (6) 441-458.
Ketchen, D. J., & James G. Combs, & Craig J. Russell, & Chris Shook, 1997.
Organizational configurations and performance: A meta-analysis. Academy of Management Journal, vol. 40, No. 1, 223-240.
Ludwig Von Bertalanffy, 1972. The history and status of general systems theory.
Academy of Management Journal 15, 407-426.
Merton, R. K. 1967. On theoretical sociology. New York: Free Press.
Meyer, A. D., Tsui, A. S., & Hinings, C. R. 1993. Configurational approach to organizational analysis. Academy of Management Journal, 36: 1175-1195.
Miles, R. E., C. C. Snow. 1978. Organizational Strategy, Structure, and Process.
McGraw-Hill, New York.
Miller, D., & Peter H. Friesen, 1978. Archetypes of strategy formulation. Management Science, vol. 24, No. 9, 921-933.
Miller, D., & Mintzberg, H. 1984. The case for configuration. Organizations: A quantum view: 125-170.
Miller, D. 1987. The genesis of configuration. Journal of Management Studies, 12:
686-701.
Miller, D. & Shamsie, Jamal, 1996. The resource-based view of the firm in two environments: The Holleywood film studios from 1936 to 1965. Academy of Management Journal, vol. 39, No. 3, 519-543.
Mintzberg, Henry, 1990. The Design School: Reconsidering the basic premises of strategic management. Strategic Management Journal, vol. 11, 171-195.
Nadler, D. A., & Tushman, M. L. 1988. Strategic organization design. Glenwood, IL:
Scott, Foresman.
Pennings, J. 1992. Structural contingency theory: A reappraisal. Research in Organizational Behavior, 14: 267-309.
Punj, G., D. W. Stewart, 1983. Cluster analysis in marketing research: Review and suggestions for application. Journal of Marketing Research, 20, 134-148.
Scott, W. R. 1981. Organization: Rational, natural and open systems. Englewood Cliffs, NJ: Prentice-Hall.
Tang, Ying-chen, & Liou Fen-May, 2007. Asset-Light Business Model: A theoretical framework for sustained competitive advantage.
Tushman, M. L., & Nadler, D. A. 1978. Information-processing: An integrating concept in organization design. Academy of Management Review 3, 613-624.
Van de Ven, A. H., & Drazin, R.1985. The concept of fit in contingency theory. Research in organizational behaviour, vol. 7: 333-365.
Weber, M. 1947. The theory of social and economic organization. New York: Oxford University Press.
Wernerfelt, B., 1984. A resource-based view of the firm’s. Strategic Management Journal, vol. 5, No. 2, 171-180.