Chiao Da Management Review 均1.32No. 1,2012 pp.1-28
開放式商業模式:積體電路產業實詮
Open Business Model:
An Empirical
Study of
IC
Industry
陳佳誰 J Chia-Yi Chen
國立屏采科技大學企業管理系
Department of Business Administration, National Pingtung University of Science and Technology
朱博湧 Po-YungChu 國立交通大學管理科學系
Department of Management Scienc巴，National Chiao Tung University 黃基鴻 Chi-HungHuang
國立交通大學管理科學系
Department of Management Science, National Chiao Tung University
摘要:在積體電路 (IntegratedCircuit ' IC)產業中，同時存在著以整合元件製
造商為主體的封閉式的商業模式，以及由虛擬整合廠商組成的開放式商業模 式 。 採行何種商業模式較能使企業獲得較高報酬或承擔較低風險是值得研究 的議題 。 本研究之目的在於檢視 IC 公司的發利報酬與其商業模式之間的關 餘，以在美國 NASDAQ 上市之IC 公司作為研究標的 ， 蒐集其 2000-2007 年間 的財務數據，並使用 Fama-French三因子模型進行資料分析。研究結果發現， 虛擬整合公司的報酬顯著優於垂直整合公司，但承擔的風險亦較高;此外， 若考慮景氣循環因素，進一步將資料分為 2000-2003 及 2004-2007 兩期間予 以比較，顯示 2004 年後虛擬登合公司的風險逐步下降，但垂直整合公司的風 險卻反而增加。此發現提供了一個有價值的觀點，說明虛擬整合已成為趨勢， 在成長的產業中將有更多公司採用開放式的商業模式 。 本文並運用各產業之 案例，闡釋如何運用及一般化本研究之實證結果 。 關鍵詞:虛擬整合;垂直整合;積體電路產業; Fama-French 三因子模型;開 放式商業模式
1 Corresponding author: Department ofBusiness Administration, National Pingtung University of Science and Technology County, Pingtung, Taiwan, E-mail: [email protected]
2 Open Business Model : An Empirical Stz呻 o/IClndustry
Abstract: In the Integrated Circuit (IC) industry, the c10sed business model which are composed of integrated device manufacturers (IDM) and the open business model which consists of virtual integrated (VI) manufactures are prevalent and both can be well justified. Whether the IDM or the VI model would better
position the firms in terms of providing a higher retum or less risk in the IC industry remains to be explored. This study aims to examine the relationship between the profitability of IC companies and their business models (IDM versus
VI). NASDAQ listed IC companies were selected as research subjects. The data were collected for the period 2000-2007 and analyzed by using the Fama-French three-factor model. The results show that VI companies significantly outperform IDM counte中arts; however, they also take higher risks than IDM firms. In
addition, a comparison of data between the periods of 2000-2003 and 2004-2007 reveals the risks of IDM firms increased while that of VI firms decreased after 2004. This finding provides a valuable insight that business model of virtual integration has been emerging as a trend and more companies have been adopting open business model in many growing industries. Events within different industries are utilized to generalize the empirical results
Keywords: Virtual integration; Vertical integration; IC industry; Fama french Three factor model; Open business model
1.
Introduction
How an industry evolves, has long been of great interest to scholars and managers. For decades, one prominent aspect of industry evolution has been the process of vertical disintegration. Vertical disintegration refers to the emergence of new intermediate markets that divide an integrated production process between two sets of specialized firms in the same industry (Jacobides, 2004). It has been ascertained that the industry life cyc1e has an impact on the degree of vertical integration of an industry. Stigler (1951) proposed that firms tend to be highly
Chiao Da Management Review Vol. 32 No. 1,2012 3
vertical disintegrated strategy in order to achieve economies of specialization and division of labour. The disintegration process of the personal computer industry is a prevailing example (Christensen et al., 2001; Christensen et al., 2002). Based on the innovation process perspective, Chesbrough (2006, 2007) proposed the concept of
“
open business model". Firms which adopt the cIassical cIosed business model are highly vertical integrated and responsible for carrying out all phases of the innovation process. In open business model, however, firrns exploit both intemal and extemal sources of innovation, while maximizing the retums that accrue 企omboth sources. Chesbrough (2006, 2007) suggested that an industry paradigm shift 企om a cIose to an open business model. Similar trends have also been found in the integrated circuit (IC) industry (Macher et al., 2002)Before 1980, the IC industry was dominated by integrated device manufacturing (IDM) and system companies which adopting traditional cIosed business model. An IDM firm is a highly vertical integrated company which incIudes IC design, fabrication, assembly, and test - all in one company. Thus, an IDM firrn not only has to allocate resources to the research and development of products, but also has to build expensive manufacturing/factory facilities. Typical IDM firms, such as Intel, AMD, Siemens, Toshiba, Hitachi, NEC, and most Korean IC firrns, appear to be the most prominent in IC product lines at the leading edge oftechnology, such as DRAMs (Macher et al., 2007)
Since the 198缸，the IC industry has been undergoing a rapid transition from
an IDM-dominated industry to a vertically disintegrated structure. The emergence
of the dedicated foundries enabled small IC design houses to build their entire product portfolios without any in-house manufacturing capacity. IC design (also called fabless) companies thus focus on designing and marketing IC products, while they outsource manufacturing to foundry service companies such as Taiwan Semiconductor Manufacturing Company (TSMC) and United Microelectronic
Corporation (UMC) in Taiwan. By allying with dedicated foundries and
assembly/test companies, IC design houses started to grow as a result of the lower en訂ybarrier to the IC business. As a result, the business model of IC industry has become more open when the vertical disintegration emerges. The evolution of the global IC industry is illus
4 Open Business Model .' An Empirical S的吵。iflClndustη
Figure 1
Evolution of Global IC Industrial Infrastructure
Sy.tem Spe個 ICD喝悟" IC F.brlcatlon ICA..embly ICTI回t
System Emerges 10M Emerg揖Fabless Foundry Emerg目 IP Vender Emerges
面回國國
同
System I I 10M
HJ~
-1970Source: lndustrial Technology lnfo!TI1ation Services ofthe Minislry ofEconomic Affairs (MOEA)
Although the IC industry has been moving from a vertically integrated to a vertically disintegrated structure
,
two different business model are prevalent and both can be well justified today. The first system is the traditional closed business model, composed of highly vertical integrated IDM firms. The other system is the open business model, which in contrast to the closed model, consists ofindependent specialized IC companies, i.e. IC design, IC fabrication, IC
assembly/test, etc. Each of the companies concentrates on specific items needed for overall IC production and they are virtually integrated to form a complete
value chain. Some scholars described such business model in IC industry as
“virtual integration" (Chu, et al., 2005).
In recent years, the virtual integration (VI) business model's share in the
global IC indus甘y has grown significantly. Several empirical studies have
indicated that firms ofVI model achieve higher returns than IDM companies (Chu
Chiao Da Management Re叫即vVol. 32 No. /.20/2
profitable strategy for IC companies. However, most of the existing literature
cannot explain why IDM and VI models coexist ifthe VI is a superior production
system in the IC industry. lt is stiU debated whether the IDM or the VI model can truly better position a firm in terms of providing higher returns or lower risks in the IC industry. Therefore, this study aims to examine the relationship between the profitability (stock retum and risk) ofIC companies and their business models
(IDM versus VI) in order to close this research gap. We propose that IDM companies achieved lower risk than firms of VI model, and therefore still hold some advantages in the IC industry. Based on the findings, this paper also attempts to predict the possible trends ofthe IC industry in the future
In the following section, we first review germane literature and formulate hypotheses. Whereafter, we illustrate the research methods and procedures. We then present our results and test the formulated hypotheses. Last, managerial implications and limitations of this study are discussed.
2. Literature Review
2.
1. Vertical Integration and Risk ReductionPrevious studies on the relationship between vertical integration and risk reduction have reported conflicting views (Krickx, 2000). The main reason for the
controversial results may be hinged on the fact that sources or types of uncertainty
were different from one study to another (Sutcliffe and Zahe仗， 1998). The relationship between vertical integration and risk reduction was first discussed by using the transactional cost theory. Literature on transaction cost theory defined uncertainty as an increase in the likelihood of opportunistic behaviour by partners,
thereby increasing transaction costs of exchange (Williamson, 1975, 1985).
Conditions of high uncertainty make it difficult for firms to draw up a complete contract (Teece, 1986). If transaction cost and Ullcertainty is high, there is strong
incentive for firms to substitute intemal organization for market exchange. Thus,
transaction cost theory proposed that vertical integration is a response to the problem of high uncertainty in order to limit the extent of potential opportunism
6 Open Business Mode/ : An Empirica/ Study ollC lndustη
by the partners. (Klein et a/., 1978; Williamson, 1975, 1985, 1991). John and Weitz (1988), Anderson (1985), and Walker and Weber (1984, 1987) provided empirical evidence for the proposition that vertical integration is an efficient solution to behaviour uncertainties. Helfat and Teece (1987) examined U.S.位rms
involved in vertical mergers between 1948 and 1979, and found a significant reduction in the systematic risk of the firm following vertical merger transactions. Chatterjee et al. (1992) also confirmed that vertical mergers are effective in reducing systematic risk, particularJy when the acquiring firm competes in a
concen甘ated market. As a result, this stream of research suggested that vertical
integration could reduce a firm's exposure to environmental uncertain旬， resulting in a lower risk for the firm.
In contrast to the prediction of transaction cost theory, some scholars
presented an altemative argument that high levels of technological uncertainty may act in the opposite direction with regard to vertical integration. For example
,
Balak.rishnan and Wemerfelt (1986) proposed a negative relationship between technological uncertainty and vertical integration, especially if the degree of the competition is high. They pointed out that fmns would rather have markets handle a technology that could quickly be obsolete than committing to it through vertical
integration. From a strategic management perspective, firms facing rapid
technological change require greater flexibility (Harrig妞， 1984, 1985; Porter,
1980) which would induce the formation of informal forms of cooperation (e.g
non-equity agreements) in order to preserve strategic flexibility. In sum, these lines of reasoning suggest that high technological uncertainty results in a decreased rather than an increased degree of vertical integration (Henderson and
Clark, 1990; Robertson and Gatignon, 1998). Several studies have a1so provided empirica1 support for the proposition that uncertainty associated with technology is likely to discourage vertical integration. For example, Heide and John (1990) and Walker and Weber (1984) found that technological unpredictability (detìned
as the inability to accurately forecast the technical requirements in the relationship) significantly reduces the expectation of continuity in buyer-supplier relationships
A possible reason of these contradictory findings in the Iiterature is that diffe
Chiao Da Management Review Vol. 32 No. 1.2012 7
vertical integration (Krickx, 2000; Sutc1iffe and Zahe仗， 1998). In order to solve this inconsistency, different types of uncertainty are introduced by researchers in order to c1arify the nature of the relation between vertical integration and risk reduction. Sutcliffe and Zaheer (1998), for example, distinguished among three main sources of uncertainty: primary uncertainty, competitive uncertainty, and
supplier uncertainty. Prirnary uncertainty refers to a lack of knowledge about states of nature
,
while competitive uncertainty is defined as the uncertainty arising from the actions of potential or actual competitors. Additionally, supplier uncertainty is the behavioural uncertainty arising from the actions of the exchangepartner firms. The authors found that primary and competitive uncertainty were negatively associated with the decision to vertically integrate, but supplier
unce此ainty was positively related to the vertical integration decision. Moreover, Chesbrough and Teece (1996) identified two types of technological innovation, namely autonomous innovation and systemic innovation. Autonomous innovation refers to innovations that can be pursued independently of others. In contrast,
systemic innovation can be realized only in co月 unction with complementary innovations. Chesbrough and Teece (1996) suggest that when an innovation is autonomous, a decentralized virtual organization can manage development and commercialization tasks quite well. However, an integrated company may be able to resolve the challenges in information sharing and coordinating for a systemic
mnovatlOn
2.2. Fama-French Three Factor Model
The Capital Asset Pricing Model (CAPM), developed by Sharpe (1964), and Lintner (1965), provides a framework in predictions for equilibrium expected returns on risky assets. Specifically, it states that the expected excess return over the risk-企eeinterest rate of an asset equals a coefficient, times the (mean-variance
efficient; market ß) market portfolio's expected excess return over the risk-free interest rate (as illustrated in equation 1). The CAP扎1: has been widely used in applications, such as estimating the cost of capital for firms and evaluating the performance ofmanaged portfolios (Fama and French, 2004)
8 Open Business Model : An Empirical Stu砂 olICIndustry
E (Rit) - Rft = ßiM [Rt - Rftl ( 1 )
Rit : return of a portfolio at time t
Rft: the risk-企eeinterest rate at time t
Rt : return of market portfolio at time t
The CAPM hypothesizes that the expected return of a sec叮itydepends only on the sensitivity of its return to the market return (market ß). Empirical evidence has shown, however, that market
ß
does not suffice to describe the expected return of a security. Other factors that seem to signi日 cantlyadd to the explanations of a security's expected return provided by marketß
inc1ude a firm's market capitalization (Ba凹， 1981; Keim, 1983) and book-to-market ratio (Fama and French, 1992, 1996a). Accordingly, Fama and French (1993, 1996b) added size and value factors to the market risk factor in the CAPM to provide a better description of expected returns, and this is known as the Fama-French three factormodel. This model posits a relationship between a security's expected return and its risk can be measured by its expos叮e to three risk factors: market factor, size factor (the return difference between small and large firms), and book-to-market ratio factor (the retum difference between high book-to-market ratio fmns and low book-to-market ratio firms). The Fama-French t趾ee factor model is illustrated in equation 2:
E(Rit) = Rft + 戶iM[Rt - Rftl +βiS岫 5MBt+ßiH.必 HMLt (2)
5MB: the return difference between small and large firms
HML: the return difference between high book-to-market ratio and low book-to-market ratio firms
The Fama-French three factor model evidently does a good job of explaining and predicting the variation of stock returns (Arshanapalli et al., 1998; Fama and French, 1996b). Therefore, this model has been used in a number of studies to examine the influence of mergers and acquisitions (Maheswaran and Yeoh, 2005) 。r brand value (Madden et al., 2006) on a fmn's stock return and risk. The
Chiao Da Managemenl Reνiew Vol. 32 No. 1,2012 9
Fama-French three factor model is eXplained in detail in Fama and French (1992,
1993).
2.3. Research Hypotheses
This section contains research hypotheses based on existing literature and research reports. In general, IC companies that adopt the VI model concentrated on a specl日c segment of the value chain that fits with firms' capabilities. For IC manufacturers (e.g. foundries， 的sembly/test companies), specializing in production can exploit scale economies, reduce costs and expand the range of potential end-user applications for semiconductors (Macher et al., 2007). The emergence of independent IC foundry and assembly/test companies also allows small IC design houses to build their product lines without substantial capital investments for intemal production capacities. Accordingly, fabless IC fmns are able to focus on IC design and product development to achieve higher retums.
Through such strategic alliances, those specialized IC companies create competitive advantages to achieve higher retums than IDM firms. In addition, IC companies today have matured to the extent that they are more likely to face autonomous, rather than systemic, technological innovation. According to Chesbrough and Teece's (1996) assumption, a decentralized vi巾alorganization could manage development and commercialization tasks quite well when an innovation is autonomous. Thus, virtuallY integrated organizations are expected to be more profitable than their IDM counte中arts
According to IC Insights, the sales contribution of fabless IC companies was less than 10% of the total sales of the semiconductor industry before 2000. In 2006, the market share of fabless firms in the IC industry increased to 20%, and IC Insights further predicted that this share is likely to increase to more than 25%
in 2011 (McGrath, 2007). These market reports suggest that the virtual integration model is growing in popularity. As for the profits of companies adopting virtual integration models, the revenues of IC designers achieved an annual growth of 16% in 2006, compared to a 9% growth for the entire IC industry. Sales 企omthe fabless semiconductor market have grown at a compound annual growth rate (CAGR) of 25% from 1998 through 2004, compared to a 9% CAGR for the entire
10 Open Business Model : An Empirical S的。今olICIndustry
IC industry (McGrath, 2007). Moreover, Chu et al. (2005) found that finns adopting the virtually integration strategy have enjoyed higher ROA and ROE than finns in the IDM group in Taiwan's IC industry. The evidence further supports the fact that the revenue of finns adopting virtual integration were continuously improving. Therefore, we hypothesize that:
Hl: The return
0/
virtually integrated firms is sign伊cantlyhigher than that0/
IDM firms.Inconsistent conclusions have emerged from previous studies conceming the relationship between vertical integration and corporate risk. Therefore, the impact of adopting the IDM or the virtual integration model on a finn's risk cannot be deduced. From an industry life cycle perspective, the IC industry is reaching its maturity stage. Thus, technological uncertainty in the IC indu甜y today is less likely to be attributed to new inventions or discoveries. Williamson (1985) proposed behavioural uncertainty as the main 釘iver of vertical integration decisions. This study thus proposes that technological uncertainty does not seem to be the critical factor influencing an IC finn's vertical integration decision Instead
,
the risks of IC finns are more likely to be determined by demand and behavioural uncertainties of partner firms. Therefore, the impact of vertical integration on an IC firm's risk may best be explained from the transaction cost theory perspective.According to the transaction cost theory, vertical integration is a solution to the problem of high asset specificity and behavioural uncertainty (Williamson,
1979). Masten, et al. (1991) argued that firms should integrate vertically to take advantage of reduced internal organization costs. In general, vertical integration allows firms to avoid an increase in the cost of market exchanges that arise from uncertainty, asset specificity and oppo叮叮lÌsm. Hence, vertical integration can be utilized by large IC firms as a viable means to reduce their risk. In other words,
firms adopting the VI model in the IC industry may face higher risks than vertically integrated IDM companies.
Moreover, virtually integrated compani巴s general1y focus on a certain segment of the value chain of IC products. Thus, firms of VI models usually
Chiao Da Managemenl Reνiew Vol. 32 No. 1.2012 11
provide narrower ranges of products or services compared to IDM companies From a diversification point ofview, firms using the VI model are more sensitive to changes in the economic cyc1e than highly integrated and diversified IDM firms. As a result, virtually integrated companies may face higher market risks than IDM companies. Based on previous discussions, it is reasonable to expect that vertical integration will help reduce corporate risk in the IC industry, reflected in the volatility of stock price. Therefore, we hypothesize
H2: The risk of virtually integrated 戶'rmsis s裙nificantlyhigher than that of IDM firms.
3. Research Methodology
3.
1. Research SubjectsIDM and VI business models have their own distinct advantages and are both prevalent in today's global IC industry. However, neither model has proven
superior. Therefore, it seems insightful to explore whether the IDM or the VI can better position IC companies in terms ofproviding higher retums or less risk. This
study attempts to examine this issue 仕omthe perspective of equity investment. If an investor intends to profit 企omthe entire IC industry, he or she can invest the
shares of an IDM company. Another option would be for an investor to own
shares in IC design houses, IC foundries, IC assembly/test companies, and other specialized IC companies, which implies that the investor has virtually owned shares of an IDM company. Assuming that the stock prices of a firm fairly and efficiently reflect its fmancial performance, the difference in profitability between IDM and VI business models could be revealed by comparing the retum and risk ofthese two investment strategies.
In this study, IC companies whose stocks or ADRs traded on the U.S. stock exchanges at any time between January 1, 2000 and December 31, 2007 are
selected as research subjects (as listed in Appendix A). These subject firms are further categorized into two subgroups, namely the IDM subgroup and the VI
12 Open Business Model : An Empirical Stu吵 o/ICJndustry
models is based on the method suggested in the Semiconductors Jndustry Annual
Report published by the Ministry of Economic Affairs (MOEA), Taiwan
(Ministry of Economic Affairs 2002), which has been cited in previous studies
(Chu et 叫，2005). More speci日cally，the IDM subgroup contains IDM companies
in the IC industry. Firms that adopting the VI model, inc1uding the independent
IC design houses, foundry service firms, assembly firms, and test firms, are
categorized as the VI subgroup.
3.2.
Research MethodologyFor the purpose of our research, we created two portfolios:
1. IDM portfolio, which is composed of firms adopting the IDM business
model
2. VI portfolio
,
which is composed of firms adopting the VI business model.These two portfolios are value-weighted and re-balanced monthly. More
specifically, the weight of each company in either the IDM or the VI portfolio is
given by the company's market capitalization (market value of all outstanding
common stock) relative to the market capitalization of all the companies in the
portfolio. These weights are recalculated at the end of each month. The financial
data used in this study are obtained from the Center for Research in Security
Prices (CRSP)
After data collection, the Fama-French three factor model was used to assess
the retums and risks of IDM and VI portfolios. The results of retums and risks of
these two portfolios were then compared in order to explore the impact of
business models on fmn performance in the IC industry. The Fama-French three
factor model posits a relationship between a security's expected retum and its risk,
which can be measured by its exposure to three risk factors: (1) market factor
(R咽Rj): market retum minus risk-free retum; (2) size factor (SMB): the retum
difference between small and large firms; and (3) book-to-market factor (HML):
the return difference between high book-to-market ratio firms and low
book-to-market ratio firms. The model used in this study can be illustrated using
Chiao Da Managemenl Reν的惜IVol. 32 No. 1,2012 13
Rit- Rft =的+ßiM [Rt - Rft] +βiSMBSMBt +βiHA在風I1Lt + Git (3)
Rit : retum ofportfolio at time t Rft: the risk-free interest rate at time t Rt : retum ofmarket portfolio at time t
5MB: the retum difference betweeo small aod large firrns at time t
HML: the retum differeoce between high book-to-market ratio firrns aod 10w
book司to-marketratio firrns at time t
In this study, the NASDAQ index was used to represent the market portfolio. In additioo, one-month Treasury Bill rate proxies the risk-企eemterest rate.
According to Chan et al. (2001) and Madden et al. (2006), two parameters from the Fama-French regression that are ofparticular importance in diagnosing a portfolio's perforrnance are the intercept terrn (叫 and the market beta (ßiM). A
positive αil indicates that a portfo1io has outperforrned its risk-adjusted beochmark; a negatJveαil iodicates underperforrnance compared with the benchmark. The market beta (ßiM) 的timates the market risk associated with a portfolio. Market beta equals to 1 indicates that the risk of the portfolio perforrns as expected. Market betas 1ess (or greater) than 1 indicate that the portfolio performs with less (or more) risk thao expected
4.
Results4.
1. Descriptive StatisticsIn this section, we first provide some descriptive statistics on the lDM, VI, and market portfolios. There are 96 monthly retums on the portfolios. The mean mooth1y retum 00 the lDM portfolio is 0.53 per cent while, during the same time period, the VI portfolio on average retumed 1.19 per cent. As predicted, the average retum on V1 portfolios is greater than that of lDM portfolios. In comparisoo, retums 00 the NASDAQ, which proxies the market return, averaged
14 Open Business Model : An Empirical Stu吵。ifIC Industry
The graph in figure 2 illus仕的es the cumulative value of IDM, VI, and market portfolios. If $1,000 was invested in January 2000, the VI portfolio yielded $1,580.57 by the end of 2007, which was higher than the 叭，120.93
generated by the IDM portfolio and $651.78 created by the NASDAQ index These results show that IC companies that adopt the VI model create more value than the IDM and market portfolio. Although IDM companies create less value than virtually integrated companies, the cumulative returns of IDM firms are still higher than the market average
Figure 2
Monthly Returns Comparison for Portfolios
1800.00 1600.00 1400.00 1200.00 1000.00 800.00 600.00 400.00 200.00 口。。 、(\、(\ .~" ~、、'、、(\、(\、司、司
JJJJJJJ J/JJJJJJJ
九九、~九九 '\.v '1:九屯九九九九九 t4.2.
ResuIts of Fama-French AnalysisThe descriptive analysis on cumulative returns described in the above section does not take risk into account. In the following, we employ the Fama-French three-factor model to adjust for risk to analyze returns on each portfolio. The Fama-French regression results ofthe two portfolios appear in Table 1.
Chiao Da Management Reνiew Vol. 32 No. 1, 2012
Table 1
The Fama-French Regressions Results
Portfolio a P P.且MB PHML IDM VI 0.30 1.59** Note: *p<0.05 ; ** p<O.OI 0.85* 1.36 0.17** -0.09 -0.22** 0.14*
The results ofthe Fama-French analysis are discussed as follows: (1) Ri sk-a句 ustedretums
15
Although the αcoefficient of the IDM and VI portfolios were both positiv巴，
only the αcoefficient of the VI portfolio was significantly different from O. The results suggested that virtually all companies who use the VI model exhibit significantly higher risk-adjusted retums than that of IDM models and the market portfolio. However, the risk-a句 usted retums on IDM companies do not significantly outperform the market average
(2) Market Beta
The IDM portfolio displayed below-average market risk (market ß=.851), which is significantly different 企om the standard benchmark value of 1 Conversely, the market
ß
of the VI portfolio is not significantly greater than or less than 1, suggesting that share price volatility of virtually integrated IC companies is not significantly different 企omthe market risk(3) Size factor
The ßSMB of the IDM portfolio was significantly greater than 0, which suggested that IDM companies exhibited a
“
reverse size effect". This finding indicated that the larger a company was, the higher retums the market expected for IDM portfolio. In co甜甜， the ßs岫 ofthe VI portfolio was not significantly different from 0, suggesting there is no size effect for VI portfolio(4) Book-to-market factor
The ßHML of the IDM portfolio was significantly less than 0, indicating that
the portfolio consisted of more companies with lower book-to-market ratios. On the other hand, the ßHML of the VI portfolio was not significantly different from 0, suggesting that there is no book-to-market effect for VI portfolio
16 Open Business Mode/ : An Empirica/ Stu砂 ollClndustry
In order to test the hypothesis of this study, independent sample t-tests were used to examine whether αcoefficients and market
ß
were significantly different between IDM and VI portfolios. H 1 predicts that the return of virtually integrated firms is significantly higher than that of IDM firms. The result of the t-test showed thatαcoefficient ofthe IDM portfolio was significantly lower than that of the VI portfolio (t=-16.84, p<O.O 1), indicating that virtually integrated IC companies achieved higher abnormal returns than IDM companies. Therefore, Hl was supported.Additionally, H2 of this study states that the risk of virtually integrated firms is significantly higher than that of IDM firms. As expected, the results of the t-test showed that the market
ß
of the IDM portfolio was significantly lower than that of VI portfolio (t=-49.01, p<O.OI). This suggested that the IDM companies ca叮y with lower risk than virtually integrated companies, which is consistent with the expectation and supports H24.3.
The Impact of Industry Life CycleSince 2004, the growth rate of the global IC industry has slowed down. Accordingly, the IC industry seems to have reached its ma個n句 stageafter 2004 Whether there are changes to the behaviour of the risks and reωrns of IC companies adopting different business models, one needs further investigation. In the following, we thus split the research period into two sections: (a) January 1, 2000 to December 31,2003, and (b) January 1,2004 to December 31, 2007. The changes in risks and returns of IDM and VI portfolios were further examined in order to provide more insight into how specifically the indus的I life cyc1e inf1uences returns and risks in IC companies.
The results of the Fama-French regression of two research sections were illustrated in Table 2. A comparison of data between the periods of January 1, 2000 to December 訓， 2003 and January 1, 2004 to December 31, 2007 showed that 也e risk-adjusted returns of the IDM portfolio and the VI portfolio both dec1ined after 2004. These results echoed the data in fig叮e 3, revealing a downturn in the IC industry. In addition, the risk-adjusted return of the VI portfolio still remained greater than that of the IDM portfolio, both pre and post
Chiao Da Managemenl Reνiew Vol. 32 No. 1,2012 17
2004. This indicates that virtually integrated companies produce a higher return
than IDM firms in the long mn. In terms of risk, the market
ß
of the IDMportfolio was significantly lower than that of the VI portfolio, both pre and post
2004, similar to the situation without considering the factor of industry cycle
Although the result again confirms that IDM companies are exposed to lower risk
than virtually integrated companies, a comparison on risk before and after 2004
revealed that the risk of the IDM portfolio increased while the risk of the VI
portfolio decreased. This finding suggests that the advantage of risk reduction for
IDM firrns seems to be diminishing
Table 2
The Fama-French Regressions Results (Time Separated)
Portfolio α p PSM8 PHML 10M (2000-2003) 1.0064 0.8438 0.2058* -0.2656** 10M (2004-2007) -0.3599 。 9606 0.0274 -0.0734 VI (2000-2003) 3.3444** 1.4000 -0.1622 。 1453 VI (2004-2007) 0.1214 1.3074 -0.1423 0.0137 Note: *p<0.05 ; ** p<O.OI
5. Discussion and Conclusion
5.
1. DiscussionsIn this study, we found that the risk-a吋 ustedreturn of the VI portfolio was significantly greater than that of the IDM portfolio. At the same time, the results
of this study revealed that the risk of stock returns of the IDM portfolio was
significantly lower than that of the Vl portfolio. Basing on these findings, we
suggest that there is a trade-off between risk and return when deciding whether to
adopt the IDM or the VI business models in the lC industry. Under the assumption of efficient markets, we may infer that although IDM companies produce
relatively lower returns compared to firms adopting the VI model, they are also exposed to lower risks. In contrast, IC companies based on the VI model may be
18 Open Business Model : An Empirical Stu砂 ofIClndus旬，
able to achieve higher returns, but they are also suffered 台omhigher risks. We have investigated the risks and returns within the IC industry prior to and after 2004 by further considering the effect of industry cycle. Although the risk-adjusted returns of the IDM and VI portfolios have both decreased due to a downturn in the global IC industry since 2004, the VI portfolio achieved higher risk-a句ustedretums than the IDM portfolio as predicted, both pre and post 2004. Therefore, we may conclude that firms adopting the VI model 叫 oy higher financial returns than IDM companies in general. In terms of market risk, the results ofthis study showed that the VI portfolio is exposed to higher risk than the IDM portfolio, both pre and post 2004. This finding indicates that there is a long-term risk advantage for IDM firms compared to firms adopting the VI model within the IC indus甘y. The risk-retum 甘ade-offbetween IDM and VI business models seems to be robust in the IC industry.
After comparing the changes in the risks ofthe IDM and VI portfolios before and after 2004, however, we did find that the risks ofthe IDM portfolio increased significantly since 2004 while the risks of the VI portfolio decreased. These results may be attributed to the strengthened strategic alliances and technological collaborations among virtually integrated companies after years of collaborating expenence . they have facilitated information and facilities sharing. Growing collaborative networks among specialized IC companies have greatly reduced the costs of transactions between contracted partners. Hence, the market risk of the VI model has been reduced substantially in recent years. This finding provides a possible explanation for the increasing enormous number of specialized IC firms world-wide. On the other hand, due to the rising R&D and installation costs within the IC industry, the risk advantage of IDM companies is eroding, which is reflected in the increase of their stock risk. In recent years, IDM companies have been cutting back on capital expenditure and outsourcing manufacturing or non-core design activities to fabless or foundry companies. Some IDM companies have even transformed to a Fabless or Fablite (used to describe IC firms with low possession of semiconductor fabrication facilities) structure. These observed trends seem to support the findings of this study.
Chiao Da Managemenl Reνiew Vol. 32 No. 1,2012 19
achieved higher returns than IDM firrns, which suggests that induced collaboration has enabled fabless firrns and foundry service firrns to use less capital and to grow faster than most IDM companies. This finding provides a reasonable explanation for the fact that the proportion of production value from specialized manufacturers within the IC industry has been rising continuously. The results of this paper are also consistent with the findings of Chu et al. (2005),
which suggests the effectiveness ofthe VI mode1 to increase a firrn's profitability. Although evidence seems to support the VI model as a more profitable s甘ucture， previous studies could not explain why IDM and VI business models still coexist Our study showed that although IDM companies produce lower returns compared to firrns based on the VI model, they are also exposed to a lower risk than virtually integrated fmns. In other words, IDM companies still maintain their advantages of lower risk to compete with firrns based on the VI model. As a result,
our findings provide a possible explanation for the coexistence of IDM and VI business models in the IC industry.
The coe伍cients of size factor and the book-to-market factor provided additional insight into the nature of the IC industry.Accumulative evidence has shown that small company stocks (those with low market capitalization) generally have higher risk-a你的ted returns than large company stocks (Banz
,
1981) Smaller companies are less efficient, have high financial leverage, and have greater difficu1ty in obtaining externa1 resources (Chan and Chen, 1991). Therefore, higher returns on small firrns are expected as a compensation for such distressing risks, which is frequently referred to as the size effect. However, the positive size factor coefficient 伊5MB)of the IDM portfolio in this study suggests a “reverse size effect" of IDM firrns where the larger the size is, the higher the stock return. A potential explanation of this resu1t is that R&D expenditure and equipment investrnents of an IDM company are enorrnous. Thus, only large scale companies in the IC industry are capable of reaching economies of scale to reduce unit costs, and hence, produce a higher returnIn the early years, the construction of a 6-inch wafer required only about US$500 million. During the era of 8-inch wafers, a factory cost between US$0.5 and US$1.5 bi1lion. Now the construction of a 12-inch wafer fabrication costs
20 Open Business Model : An Empirical Study o/IC Indust.η
US$2.5 billion and the 65nm production design costs at least US$45 million. Only
a few 10M finns are able to afford such an enormous investment independently Although the use of 12-inch wafers can significantly reduce the unit cost, the increasingly unaffordable and substantially higher investment costs are the key factors that drive IDM finns to increase their outsourcing activities and strategic alliances with specialized IC companies (Fabless and Foundries). Some 10M finns are even transitioning themselves toward Fablite structur肘， which suggest
that the industry may be changing towards virtual integration in the future On the other hand, the negative book-to-market factor coefficient (ßH地) suggests that the 10M portfolio consists of companies with a low book-to-market ratio. This finding indicates that the market is still optimistic about the growth potential of IDM companies. Although there is an increased share of the VI model within the IC industry, 10M finns currently still occupy around 80% of the market. The major European, US, and Japanese 10M companies are still in possession of more comprehensive patents and advanced production technologies. Some industry observers have projected that due to their advantages in new product or technology developments and superior capabilities of independent mnovatl凹， 10M companies will still dominate the IC industry in the 臼ture (LaPedus
,
2005). Consequently,
there is still a debate over whether the 10M and VI models would grow or decline in the future.5.2.
Theoretical Implications and GeneralizationsIt is concluded that the Vl portfolio has higher return and higher risk than the 10M portfolio. This result is consistent with the common concept that high retum always goes with high risk. But, a more detailed breakdown shows the risks of the VI portfolio decreased since 2004 while the risks of the 10M portfolio increased. The results support Chesbrough's (2006, 2007) proposition that the industry paradigm is shifting to open business model. This finding provides a valuable insight that business model of virtual integration has been emerging as a trend and more companies have been adopting open business model in many growing open industries, such as IC industry, consumer electronic industry, PC industry and so on. Here, the open industry means the platfonn of collaboration
Chiao Da Management Reνiew Vol. 32 No. 1,2012 21
among participators in the industry has been operating efficiently for a long time.
Thousands of newly enterprises emerge and disappear every day. What could managers draw some implications from our empirical finding?
5.2.1. Theoretical Supports
Business model innovation is one of the most profound ways to differentiate
a business nowadays. The definition of innovation used to be synonymous with
invention. It was the realm of R&D exclusively. In the open business model, it's
about commercialization. It is about business model innovations as much as
products and processes. The integration of the business activities extends the business processes across organizations. How to innovate in the 21 st century
global economy is al\ about the ability to create value, capture value, scale up, and
integrate technologies intemal as well as extemal. Chesbrough (2006, 2007) argued that open business models foster col\aborations with customers and
suppliers to everyone's benefit. The more companies leam about open business models, the more they realize how much they have to change their own
innovation activities to take ful\ advantage of the paradigm shift. lt is not simply a
matter of searching for new technologies. To thrive, companies must adapt their business models to make them more open to extemal ideas
,
technologies andpaths to the markets
5.2.2. IC Industry Event and Evidence
It was announced on March 2, 2009 that lntel and TSMC (Taiwan Semiconductor Manufacturing Company) would col\aborate on Atom-platform chip production. In闕，which prides itself on doing chip manufacturing in-house, is lending its Atom technology including processes, IP and design f10ws to TSMC TSMC is the largest contract chip manufacturer in the world
The col\aboration between Intel and TSMC would create a win-win game.
The shift is part of Intel's effort to push its x86 architecture, which is used in the
vast majority of the wor!d's personal computers and servers, into a wider a叮ayof lower-cost consumer devices such as Netbook and mobile intemet devices (MID). Al\ products that emerge f全om the collaboration will be lntel-branded products
22 Open Business Mode/ : An Empirica/ Study olIC lndustry
Collaborating with TSMC, Intel could focus on its core business and leverage TSMC's production excellence to other considerable big-volume low-price market without huge capacity investment. As a result, Intel would not only reduce its risk for less capital expenditure on fab but also increase its return with new market expansion. For its part, TSMC can complement the new CPU product lines and further promote the innovative business model of IC foundry service.
The winning of Intel business would not only improve the utilization of TSMC advanced process capacities but also attract more IDM players to follow Intel's
suit. Via such TSMC could improve its return, market position as well as reduce risk. The ramifications of strategic behaviour of Intel, the leading IDM player,
and TSMC, the dominant foundry manufacturer, confirm the empirical findings of the research. This event also provides an empirical evidence for the spread of open business model in the IC industry
5.2.3. Generalizations
Recently, Sony has agreed to build a strategic alliance with Taiwan's Foxconn Electronics (Hon Hai Precision Industry Co. Ltd.) for the production of LCD TVs for the Americas region. Foxconn is the world's largest contract electronics manufacturer
,
making advance products such as the iPod and iPhone for Apple, PCs for HP and more.Under the agreement, Sony is concentrating internal resources towards areas that contribute to product differentiation, such as R&D, engineering and design,
while also establishing a structure that enables the company to bring attractive products to market at the earliest possible oppo前山lÌty. At the same time
,
Sony will outsource production of its popular TVs to Foxconn Electronics in order to reduce massive fixed asset expenditure and production cost, as well as to improve profitability and business expansionThis strategic alliance provides another empirical evidence of the adoption of virtual integration strategies in a different growing industry. Managers could find other empirical evidences of open business model in many industries such as PC
,
NotebookINetbook, telecommunication products, pharmaceuticals and Bioscience resulted from the trend of globalization and collaboration.Chiao Da Management Reνiew Vol. 32 No. 1,2012 23
5.3. Limitations and Future Research
Several limitations of this investigation should be noted. First, stock retum
and risk were used as proxy variables for a tìrm's protìtability in this study However stock retum and risk may not fully reflect the tìnancial information of a
company's retum and risk. Future studies may incorporate other tìnancial models
or methodologies to compare the performance between IDM and VI business
models. Second, because of the nature of the Fama-French three factor model,
only IC companies listed on the U.S. stock exchange were selected as research
subjects. Future work could employ other methodologies that allow for more IC
companies world-wide to be analysed in order to obtain a more comprehensive understanding of advantages and disadvantages of IDM and VI business models in the IC industry
6.
ReferencesAnderson, E. (1985), “The Salesperson as Outside Agent or Employee: A
Transaction Cost Analysis," Marketing Science, 4(3), 234-254
Arshanapalli, B., Coggin, T. D. and Doukas, J. A. (1998),“Multifactor Asset
Pricing Analysis of Intemational Value Investment Strategies," Journal of
Portfolio Management, 24(4), 10-23
Balakrishnan, S. and Wemerfelt, B. (1986),“Technical Change, Competition and
Vertical Integration," Strategic Management Journal, 7(4), 347-359.
Banz, R. W. (1981),“The Relationship between Retum and Market Value of
Common Stocks," Journal of Financial Economics, 9(1)， 于 18
Chan, K. C. and Chen, N. (1991),“Structural and Retum Characteristics of Small
and Large Firms," Journal ofFinance, 46(4),1467-1484.
Chan, L. K. c., Lakonishok, J. and Sougiannis, T. (2001),“The Stock Market
Valuation of Research and Development Expenditures," Journal of Finance,
56(6)
,
2431-2456Chatterje巴，鼠， Lubatkin， M. and Schoenecker, T. (1992),“Vertical Strategies and
24 Open Business Model : An Empirical Stu砂 olICIndustry
138-156
Chesbrough, H. W. and Teece, D. 1. (1996),“When is Virtual Virtuous? Organizing for Innovation," Harvard Business Review, 74(1)， 的 -73
Chesbrough, H. W. (2006), Open Business Models: How to Thrive in the New lnnovation Landscape, Boston, MA: Harvard University Press
Chesbrough, H. W. (2007),“Why Companies Should Have Open Business Models," MlT Sloan Management Review, 48(2), 22-28
Christensen, C. M., Raynor, M. and Verlinden, M. (2001),“Skate to Where the Money Will Be," Harvard Business Review, 79(10), 72-81
Christensen, C. M., Verlinden, M. and Westerman, G. (2002), “Disruption, Disintegration and the Dissipation of Differentiability," lndustrial and Corporate Change, 11(5),955-993.
Chu, P. Y., Teng, M. J., Huang, C. H. and Lin, H. S. (2005),“Virtual Integration and Profitability: Some Evidence from Taiwan's IC Industry," lnternational Journal ofTechnology Management, 29(1), 152-172.
Fama, E. F. and French, K. R. (1992),“The Cross-Section of Expected Stock Retums," Journal of Finance, 47(2), 427-465
Fama, E. F. and French, K. R. (1993),“Common Risk Factors in the Retums on Stocks and Bonds," Journal ofFinancial Economics, 33(1), 3-56.
Fama, E. F. and French, K. R. (1996a),“The CAPM is Wanted, Dead or Alive," Journal ofFinance, 51(5),1947-1958
Fama, E. F. and French, K. R. (1996b), “Multifactor Explanations of Asset Pricing Anomalies," Journal ofFinance, 51(1), 55-84.
Fama, E. F. and French, K. R. (2004), “The CAPM: Theory and Evidence," Journal ofEconomic Perspectives, 18(3),25-46
Harrigan, K. R. (1984),“Formulating Vertical Integration Strategies," Academy of Management Review, 9(4), 638-652.
Harrigan, K. R. (1985),“Vertical Integration and Corporate Strategy," Academy of Management Journal, 28(2),397-425
Heide, 1. B. and John, G. (1990), “Al1iances in Industrial Purchasing: The Determinants of Joint Action in Buyer-Supplier ReJationships," Journal of Marketing Research, 27(1), 24-36
Chiao Da Management Review Vol. 32 No. 1,2012 25
Helfat, C. E. and Teece, D. J. (1987),“Vertical Integration and Risk Reduction," JournalolL帥，Economics, and Organization, 3(1), 47-67
Henderson, R. M. and Clark, K. B. (1990), “Architectural Innovation: The
Reconfiguration of Existing Product Technologies and the Failure of
Established Finns," Administrative Science Quarte巾，35(1), 9-3 。
1acobid白，M. G. (2004),“Industry Change through Vertical Disintegration: How
and Why Markets Emerged in Mortgage Banking
,"
Academy01
Management Journal, 48(3), 465-498
1ohn, G. and Wei泣， B. A. (1988),“Forward Integration into Distribution: 的
Empirical Test of Transaction Cost Analysis," Journal
01
Law, Economics,and Organization, 4(2), 337-355.
Keim, D. B. (1983),“Size-related Anomalies and Stock Retum Seasonality,"
Journal olFinancial Economics, 12(1), 13-32
Klein, 8., Crawford, R. G. and A1chian, A. A. (1978),“Vertical Integration, Appropriable Rents, and the Competitive Contracting Process," Journal
01
Law and Economics, 21(2), 297-326.Krickx, G. A. (2000), “The Relationship between Uncertainty and Vertical
Integration," lnternational Journal olOrganization Ana仰的，8(3), 309-329
LaPedus, M. (2005), “Challenges Seen for Seamless DFM, Says Expert,"
available at﹒ h句://www.eetasia.com/ART_8800371275_480100_NT_a458
d9cc.HT卸1.
Lintner, J. (1965),“The Valuation of Risk Assets and the Selection of Risky
Investments in Stock Portfolios and Capital Budgets," Review
01
Economicsand Statistics, 47(1), 13-37
Macher, J., Mowery, D. C. and Minin, A. D. (2007),“The 'Non-Globalization' of
Innovation in the Semiconductor Industry," Ca/ifornia Management Review,
50(1),217-242
Macher, 1. T., Mowery, D. C. and Simco巴， T. S. (2002), “E-business and
Disintegration of the Semiconductor lndustry Value Chain," lndustry and
lnnovation, 9(3),155-181
Madden, T. J., Fehle, F. and Foumier, S. (2006),“Brands Matter: An Empirical
26 Open Business Model : An Empirical Stu砂 ollClndustry
Journal of the Academy of Marketing Sc肌肉 34(2)，224-235
Maheswaran, K. and Yeoh, S. C. (ρ2005) ，
Some Australian Evidence," Australian Journal of Management, 30(1), 111-126.
Masten, S. E., Meehan, J. W. and Snyder, E. A. (1991), “The Costs of Organization," Journal of Law, Econom悶，and Organization, 7(1),1-25 McGrath, D. (2007),“Fabless Firms Grabbed 20% of IC Sales," EE Times Asia,
available at: h句 ://www.eetaSiamm/ART-88004500271802OO-Ntoa22 be4.HTM
Ministry of Economic Affairs (2002), 2002 Semiconductors lndustη1 Annual
Report, Taipei: Minis甘yof Economic Affairs.
Porter, M. E. (1980), Competitive Strategy, New York, NY: Free Press.
Robertson, T. S. and Gatignon, H. (1998),“Technology Development Mode: A Transaction Cost Conceptualization," Strategic Management Journal, 19(6), 515-531
Sharpe, W. F. (1964),“Capital Asset Prices: A Theory of Market Equilibrium under Conditions of Risk," Journal of Finance, 19(3), 425-442.
Stigler, G. 1. (1951),“The Division of Labor is Limited by the Extent of the Market," Journal of Political Economy, 59(3), 185-193.
Sutcliffe, K. M. and Zaheer, A. (1998), “Uncertainty in the Transaction
Environment: An Empirical Test," Strategic Management Journal, 19(1),
1-23.
T閃閃，D. J. (1986),“Profiting from Technological Innovation: Implications for
Integration, Collaboration, Licensing and Public Policy," Research Policy, 15(6),285-305.
Walker, G. and Weber, D. (1984), “A Transaction Cost Approach to Make-or-Buy Decisions," Administrative Science Quarter鈔， 29(3),
373-391
Walker, G. and Weber, D. (1987), “Supplier Competition, Uncertainty and Make-or-Buy Decisions," Academy of Management Journal, 30(3), 589-596
Chiao Da Management Review Vol. 32 No. /,20/2 27
Implications: A Study in the Economics of lnternal Organization, New
York, NY: Free Press
Williamson, O. E. (1979),“Transaction-Cost Economics: The Govemance of
Contractua1 Re1ations," Journal of Lαw and Economics, 22(2), 233-261
Williamson, O. E. (1985), The Economic Institutions ofCapitalism, New York,
NY: Free Press.
Williamson, O. E. (札1991)，
Discrete Structura1 Altema泓tive郎s ，"Admini衍st什'rativeScience Quarter~砂y， 36(2)，
28 Open Busin也sModel : An Empirical Sludy oflC lndustry Appendix A Portfolio VI Broadcom Portfolio IDM .國到 TI Spansion Freescale Maxim NEC Fairchild Micron On semi . 區區-祖國岫暉-四"超過曙臨 IBM Silicon Lahoratories Cypress Amkor MPS 一 一，←一一 -E 圓圓圓圓圓， )f圓圓圓圓團--﹒圓圓圓圓圓圓圈祖國恥暉， Sandisk Zoran Xilinx Cirrus Altera TS恥1C Conexant Chartered Catalvst DSP Himax Linear PLX Standard Microsvstems Vitesse E 圓圓圓圓圓團團體盤姐lIIIIdI盤組團-圖-圖自 圖IIIIIIII函圓圓圓晶祖
