The history of the standard for the calculation of cost of ownership in the semiconductor industry

Abstract

The semiconductor industry is one of very few industries to have a standard for management accounting, and this concerns a method for calculating the cost of ownership (COO). This research investigates the history of the development of the COO standard, starting from the late 1980s and stretching to the mid-1990s, and explores the circumstances under which this development occurred. We find that the development and revision activities for COO built on complementary conditions, such as industry organizations, networks of professionals, and standard-setting procedures, which had been established for cooperation in research and development and for the development of technical standards. We suggest that these factors may explain the absence of standards in management accounting in many other settings.

Keywords

cost of ownership management accounting standards semiconductor industry standard development history

Introduction

Several studies (e.g. Camfferman, 2012; Camfferman and Zeff, 2015; Coronella et al., 2013; Richardson, 2011) investigate the “historical roles, processes or central figures of accounting standard setting and advisory bodies in the quest for improved national or supranational regulation of external reporting” (Bisman, 2012: 15). While accounting standards are rather common, standards for management accounting are relatively rare. Despite early calls for standardization of cost information (Hergert and Morris, 1989; LaLonde and Pohlen, 1996), only a few management accounting standards seem to exist (Granlund, 2011).

The topic of standards for management information has been discussed in the literature, although Caglio and Ditillo (2008) identified the topic as under-researched in their review of the literature on interorganizational management accounting. While some papers describe the history of particular standardization initiatives of management accounting inside a large company (Chandar et al., 2012; Collier, 2012), the notion of management accounting standards implies use across multiple organizations. A few empirical studies provide examples of cost models that extend beyond a single firm, often initiated by a large and powerful customer who enforced this on his suppliers (Dekker, 2003; Dekker and Van Goor, 2000; Schulze et al., 2012). These cost models included common definition of activities. However, these studies do not address the issue of standardization across multiple supplier–customer relationships that companies are usually part of. Such industry-wide standardization of cost information would be important, but rarely happens (Kajüter and Kulmala, 2005; McIvor, 2001; Seal et al., 1999). With some notable exceptions such as the case of the “Blue Book” (Jack, 2015), very few studies provide empirical data about industry-wide standardization efforts.¹ The Blue Book was an attempt to standardize management accounting in the Australian agricultural sector and to develop a comprehensive and uniform code and chart of accounts (Jack, 2015).

Another of the rare examples of an established industry-wide standard for management accounting can be found in the semiconductor industry.² This standard concerns a defined method for the calculation of the cost of ownership (COO hereafter) of semiconductor manufacturing equipment (Heilala et al., 2006; Ragona, 2002).³ In general, COO refers to “the cost of equipment employed in manufacturing including the costs associated with consumables, operating and maintaining the system, and the lifetime of operational use of the system” (Trybula, 2006: 614). COO is an influential form of management accounting in this industry, and guides large investment decisions (Miller et al., 2012; Miller and O’Leary, 2007). Our research objectives are (a) to document an intriguing accounting story stretching from the late 1980s to the mid-1990s, (b) to provide insights on the circumstances why this development occurred, and (c) to reflect on what this occurrence may have to say more broadly about standard development in accounting. We believe that the E35 standard published by Semiconductor Equipment and Materials International (SEMI hereafter) and its history provide a fascinating example to study for the following three reasons that also provide the motivation for our study.

First, the semiconductor industry’s standard is the only management accounting standard for COO that is voluntary, public, and widely used (Geissdörfer et al., 2009). The SEMI E35 standard was initially published in 1995 by SEMI, which is a global industry association for companies in the micro- and nano-electronics industries. Since then, the SEMI E35 standard has been continuously revised and is described in publicly available documents (SEMI, 2012a, 2012b). In contrast, the often referred to “Gartner Total Cost of Ownership” model for information technology is not a voluntary, public standard, but is a commercial service first launched in 1987 by the technology research firm Gartner (Mckeen and Smith, 2010; Mieritz and Kirwin, 2005). The methodology of the latter is partially disclosed only to clients, and any analyses and methods are strictly for internal use by Gartner clients. In contrast, the SEMI COO standard is widely used (Dance et al., 1996) and has been implemented in software systems purchased by over 3,000 companies in the semiconductor industry.⁴ Previous literature shows the important role of COO in the semiconductor industry (Hazelton et al., 2008; Miller and O’Leary, 2007; Seidel, 2007; Silverman, 2005), and mentioned that a detailed, influential standard for these calculations existed (Geissdörfer et al., 2009). However, despite the SEMI E35 standard’s unique presence in management accounting, the history of the development of this standard has not been described in the literature. Using new information from interviews in combination with publicly available data, we disentangle the story of how the SEMI E35 standard was formed.

Second, while accounting history literature provides interesting examples of standard developments in cost accounting that failed such as the “Blue Book” (Cortese, 2011; Jack, 2015), no prior study clarifies why the unique COO method succeeded – and why it happened in the semiconductor industry. We draw on the characteristics of the semiconductor industry to provide an explanation for the existence of this exceptional standard.

Third, our explanation identifies interaction with the development of technical standards as a factor that may influence opportunities for standard development in management accounting. As far as we know, this interaction has not been discussed in the literature, which has looked at other factors that influence standard development, such as lobbying (Cortese, 2011), regulation (Levant and Nikitin, 2012), and changes in information technology (Jack, 2015). Our study suggests that connection to technical standards (in particular the organization that handles the standards) may play a vital role in the development of management accounting standards, offering a new perspective for examining the development of standards in management accounting.

Research method

The present study is based on publicly available documents, such as papers published in academic and professional journals, Internet pages, and documents that can be downloaded for free (such as industry reports and presentations) or at a moderate price (such as the standard documents). Additional information came from interviews of 17 experts on COO in the semiconductor industry (in person, via telephone or Skype, and by email), who work or have worked at industry organizations (SEMI, SEMATECH, International Technology Roadmap for Semiconductors (ITRS hereafter)), software companies (Wright Williams & Kelly, Inc. (WWK hereafter) and IC Knowledge), semiconductor equipment companies (ASML, Centrotherm, RENA), integrated-circuit (IC) manufacturing companies (Texas Instruments (TI hereafter), Infineon, NXP, Intel, AMD, Global Foundries, and others), engineering consulting firms, and other companies in the semiconductor industry, such as material suppliers. Many of these experts had been involved in standard setting for COO calculations and had conducted COO analyses for many years. The experts are listed in Appendix 1 and the interviews with them took place between 2012 and 2017. With some of the listed experts, we conducted multiple interviews or we received several emails from them to answer follow-up questions, which are also indicated in Appendix 1.

The qualitative data analysis was guided by our initial questions as to how and why the SEMI E35 standard had been developed. At first, we based our research on publicly available information, such as key papers on COO in the semiconductor industry and websites of involved parties. Initial relevant experts were then identified based on this public information. For instance, we contacted authors of COO papers as well as representatives from involved industry associations. The main topics addressed by our questions were the usage of COO and the COO standard in the semiconductor industry as well as the COO standard development process. Regarding the latter, questions focused on which events happened when and why, and who was involved. Questions were asked to let experts describe what happened and to provide specific examples for their claims. Analyzing the obtained qualitative data was a process of connecting the various pieces of information we obtained, as well as discovering gaps and inconsistencies that sparked new questions. This process led to collecting further information by asking follow-up questions to interviewees, including asking them to provide further contacts that enabled us to expand the circle of experts we talked to. This effort led to triangulating the information obtained from experts with information obtained from other experts as well as with publicly available information. As the story unfolded, the themes and questions become more nuanced and specific, and the resulting answers contributed to closing the holes in our understanding.

In this article, we will document the history based on the combination of those various sources of information. For readability and brevity purposes, the majority of the information gained from the interviews is rephrased and referenced via endnotes. The dates mentioned in these endnotes refer to when we collected the information, and the number in square brackets refers to the expert as they are listed in Appendix 1.

COO standard

We used the SEMI E35-0312 standard, which can be purchased on the SEMI website (http://www.semi.org). It is the outcome of seven revisions since its initial publication in 1995,⁵ and its purpose is “to provide standard metrics for evaluating unit production⁶ cost effectiveness of manufacturing equipment⁷ in the semiconductor related industries” (SEMI, 2012a). The guide is applicable to any type of equipment for processing semiconductor units, such as IC wafers and devices, and may also be applicable to other manufacturing equipment such as flat panel displays, photovoltaics, or hard disk drives.

The SEMI E35 splits COO into the “cost of equipment ownership” (CEO) and the “cost of yield loss” (CYL). CEO can be understood as the sum of fixed and variable costs in relation to the amount of good units produced:

C E O = \frac{(F C + R C) \times E R}{T P T \times Y \times U}

The “fixed costs per unit of equipment” (FC in the formula and hereafter) and the “recurring costs per unit of equipment” (RC in the formula and hereafter) are summed and multiplied with the “amount of equipment required” (ER in the formula and hereafter), since the “fixed costs” and “recurring costs” are measured per piece of equipment. The elements FC and RC are broken down into five “cost categories” (e.g. “equipment”) that comprise 20 “cost elements” (such as “acquisition,” “installation,” and “training” for the “equipment” cost category) before they are summed up. In the denominator, which reflects the total amount of good units⁸ produced, TPT stands for “throughput,” Y is “composite yield,”⁹ and U is “utilization” of equipment. The methodology for determining the particular “cost elements” is documented precisely. For example, for the cost element “training” within the “cost category” of “equipment,” the method is formulated as follows:

Expressed in terms of actual burdened costs for labor-hours of training multiplied by the number of training hours and training course costs, if applicable, for each equipment user’s personnel type. Note that depending on accounting rules, some training course costs may be included as part of the equipment acquisition cost and should not be double counted. (SEMI, 2012a)

The second element, CYL, expresses the “costs incurred by yield losses per good unit.” CYL reflects that a unit lost at the end of a particular manufacturing process step causes a cost equal to the cost of the starting unit plus the manufacturing cost of the step at which it was lost. For the sake of brevity, we do not discuss CYL in further detail.

The SEMI E35 guide does not include all elements for the COO calculations. Several input parameters and calculations are defined in other SEMI standards, particularly the SEMI E10 standard. These elements are described in a different, technical standard because these parameters and calculations are also needed for other purposes, such as exchanging data among different tools in a production line.

Reliability, availability, and maintainability standard (SEMI E10)

The SEMI E10 standard is called “specification for definition and measurement of equipment reliability, availability, and maintainability (RAM) and utilization” (SEMI, 2012b). The first edition of this standard was published in 1986, several years before the SEMI E35 standard describing COO.¹⁰ The SEMI E10 standard provides the definitions and calculations that underlie “total utilization” and “operational uptime,” which are key parameters in the COO calculation. Furthermore, this standard includes the definition and calculation of the “costs of consumable material, of non-consumable parts, and of maintenance,” which are also part of the COO calculation.

The core of this standard is the definition of six “operating states” of equipment that cover all equipment conditions and periods of time. The most important are (a) “non-scheduled state”: time when the equipment system is not scheduled to be used in production (e.g. holidays out of the production schedule); and (b) “unscheduled downtime state”: time when the equipment system has experienced a failure until equipment is restored to a condition where it may perform its intended function (e.g. replacing a broken component). Four other states are defined in the SEMI E10 standard but will not be discussed in detail here: (c) “scheduled downtime state,” (d) “engineering state,” (e) “standby state,” and (f) “productive state” (Oechsner et al., 2002). The operating states 2–6 are called “operations time,” consisting of “downtime” (2 and 3) and “uptime” (4–6). The “unscheduled downtime state” and the “scheduled downtime state” are each formally broken down into eight more detailed “substates,” and the SEMI E10 standard document describes several activities included in each “substate.” Also, 26 performance measures are defined on the basis of the “substates.”

Standard development

Historical context and the first COO models in the semiconductor industry

Before the 1980s, most US firms would choose suppliers on the basis of the contract’s bottom line, that is, the quoted initial purchase price of the equipment (Ellram and Siferd, 1993). Also in the semiconductor industry, initial purchase costs as well as installation costs were the main purchasing decisions factors (Trybula, 2006). Important attributes such as equipment reliability, yield, and utilization, while having a major impact on the costs during the lifetime of the equipment, were considered separately in making equipment purchasing decisions and not monetized using COO. In the 1980s, cost pressure entered the semiconductor industry with new players emerging, particularly from Japan. US semiconductor companies were under increasing pressure from emerging Japanese competition with strength in products such as dynamic random-access memory chips. In 1987, a US report found that Japanese manufacturers outperformed US companies in 12 of 25 major semiconductor products while the US companies excelled in only five (Brown and Linden, 2009). Also, Japanese firms had lower costs of capital than their US competitors, which increased pressure to remain competitive (Irwin, 1996; Shoven and Topper, 1992). Existing semiconductor manufacturers found themselves on the losing side of the competition, as new equipment and new ways of thinking were required to remain competitive.¹¹ Rising equipment costs and an increasing importance of cost per unit to manufacturers led to the development of COO (Carnes and Su, 1991).

The need for a method like COO can be illustrated with the case of two kinds of semiconductor manufacturing technologies for lithography competing in the late 1980s for IC manufacturing, namely, projection imaging and near-contact imaging (stepper). Projection imaging had lower equipment costs and higher TPT rates. However, operators and engineers of the equipment knew that the yields were lower and more maintenance was required in comparison to the stepper process. A method was needed to evaluate the long-term cost impact of the process and the semiconductor manufacturing equipment not solely on the basis of purchase price and installation costs but also on elements such as reliability, utilization, and yields (Abramson et al., 1997: 220–221).

The first COO calculation models in the semiconductor industry were independently developed during the mid- to late 1980s by several companies that included both equipment suppliers and IC manufacturers such as Intel (Lafrance and Westrate, 1993). Intel and TI each had internal groups for COO modeling, and the groups also had informal professional contacts among each other, as explained by an expert managing such a team:

I know that later Intel had a small group working specifically on developing a COO spreadsheet and methodology that corresponded to the small group at TI that I managed that worked on COO software/methodology/global implementation.¹²

As different COO models in the semiconductor industry were being developed independently in several companies, the threat of inconsistent or incomparable COO values was strong. Calculating COO was complex given how many different inputs (fixed costs, recurring costs, yield, etc.) needed to be defined. Considering the amount of material suppliers, equipment suppliers, and semiconductor manufacturing companies in the industry, having various different COO models that would need to be understood when exchanging COO information would have greatly hampered the practicality of COO to improve overall costs in the industry.¹³ For example, one expert commented as follows on the practicality of understanding each other’s COO models:

TI spent weeks trying to determine why the AMAT [Applied Materials – a semiconductor equipment manufacturer] COO model was not aligning with the TI COO model results. In the end, it was just easier to demand compliance with a standardized approach, which TI did.¹⁴

Initial development of COO models at SEMATECH

In the late 1980s, SEMATECH started working on COO modeling. “SEMATECH” stands for “semiconductor manufacturing technology,” and it is a not-for-profit consortium performing research and development (R&D) for member semiconductor manufacturers to advance chip manufacturing. It was founded in the challenging economic environment in 1987 by 14 US-based semiconductor manufacturers in cooperation with the US government to strengthen the US semiconductor industry by jointly solving common manufacturing problems.¹⁵ While the stated objective was to focus on precompetitive R&D, another unstated objective was worldwide competitive analysis to discover why existing companies were losing their competitive edge. Cost and economic models such as COO were also developed to help understand that question of competitiveness.¹⁶

One of the promoters of COO was Dean Toombs, who brought the concept to SEMATECH as an Intel assignee in the late 1980s (Dance et al., 1996; Lafrance and Westrate, 1993). Prior to his work at Intel, he was a vice president at TI, and therefore in a position to know what COO work had been done at both companies.¹⁷ He was involved in the founding of SEMATECH and became a high-level SEMATECH manager, and was instrumental in getting management support and funding directed toward cost modeling.¹⁸ This has been emphasized by an expert involved in COO modeling and standard setting since the beginning:

Dean was in a unique position of knowing what COO work had been done at TI and what was being done at Intel … When Dean was actively involved from Intel with SEMATECH in its early days, I believe he was able to be a key driver/supporter of cost of ownership methodology as one of the key equipment performance metrics of their equipment evaluation activities. Someone else could have done so, but I am not sure who that would have been or if it would have occurred.¹⁹

The SEMATECH Cost Modeling Technical Advisory Board (TAB hereafter) and the SEMATECH Cost Modeling Users Group (Users Group) were formed, comprising specialists from various IC companies. Ross Carnes, a Motorola assignee at SEMATECH, took over the COO effort (Abramson et al., 1997). Daren Dance, a SEMATECH employee, was one of the key experts, as was Charles Stapper, an IBM employee not assigned to SEMATECH. Both published on semiconductor yield modeling, which was important for the SEMATECH COO model.²⁰ Another expert was Rick Jarvis, an assignee from AT&T/Lucent (1985–1990), primarily working together with Daren Dance on yield modeling in that period. With the ideas of the people in the TAB and Users Group as a foundation, COO was developed at SEMATECH as a sophisticated spreadsheet model around 1989.²¹ In 1991, Ross Carnes published on COO and announced intentions to develop the presented model into an industry standard. For this model, equipment state definitions from the SEMI E10-90 standard were used for determining utilization (Carnes and Su, 1991). Later in 1991, Daren Dance became responsible for the COO effort at SEMATECH (Abramson et al., 1997). The developed SEMATECH COO spreadsheet model was initially exclusively distributed to SEMATECH and SEMI/SEMATECH members (Lafrance and Westrate, 1993).²²

At the insistence of the TAB, this spreadsheet COO software was made publicly available to the entire industry. For it to be most useful, both the equipment suppliers and the users (i.e. SEMATECH member companies) needed to be using the exact same COO calculations to share and validate the results. This practice quickly led to its general adoption by the semiconductor industry as the “de facto” industry standard COO spreadsheet, at least partially because it was more comprehensive that what most individual companies were using.²³ By July 1993, more than 600 copies of the SEMATECH COO model had been distributed, and other industry associations were promoting the use of COO (Lafrance and Westrate, 1993).

With a broader user base, the problem of user-friendliness arose, which resulted in improved versions of the spreadsheet model. However, functional improvements made the model more complex.²⁴ SEMATECH recognized such problems and responded with a series of training programs (Lafrance and Westrate, 1993). In spite of this effort, the problems could not be solved entirely, and the TAB insisted that SEMATECH should either make significant improvements, especially in the area of user friendliness,²⁵ and provide ongoing technical support and training, or work with an outside company to commercialize the model.²⁶ Some of the desired changes of the SEMATECH spreadsheets were the addition of a database for the inputs for a specific scenario to be stored, retrieved, and reused rather than saving each COO scenario in the full spreadsheet, and the availability of default values.²⁷ The option to manage the improvements and trainings internally would have been outside of the organization’s scope, and at that time, SEMATECH, a US government-funded organization, was banned from supporting non-US suppliers. Also, SEMATECH had no policies for developing, supporting, and publishing software and standards. A July 1993 publication aimed at raising awareness of COO analysis communicated the intention to provide a commercial software package in the future and to standardize COO (Lafrance and Westrate, 1993).

Transfer to WWK and SEMI

The following subsections deal with two relevant historical processes that happened between 1993 and 1995 which partially overlapped: the commercialization of the SEMATECH COO model as COOL and TWO COOL, followed by the standardization of COO jointly with SEMI, leading to the generation of the SEMI E35 standard. We first present the commercialization of the SEMATECH COO software because this predated the standardization of the COO method jointly with SEMI.²⁸

Commercialization with WWK

A request for proposals to take over and further develop the existing COO model was put out in 1993.²⁹ Daren Dance led an open bidding process to find a partner for commercialization of the SEMATECH COO model as a professional software package (Abramson et al., 1997). A SEMATECH expert involved during the bidding process provided further details:

Bidders included consulting companies with which SEMATECH had working relationships (such as WWK), SEMATECH member companies, semiconductor equipment suppliers, and individuals associated with universities with which SEMATECH had working relationships. [The received bids were reviewed and] purchasing forwarded acceptable bids to the SEMATECH modeling and simulation group where these were compared both technically and commercially with a previously developed matrix. The top bids were presented to SEMATECH executives and WWK was awarded the contract on technical completeness, experience and plans for commercial support and training.³⁰

This happened in 1994. The head of the firm WWK David Jimenez had previously been familiar with the SEMATECH COO model as highlighted by his contribution of a section on “What is Cost of Ownership” in a publication from 1993 (Lafrance and Westrate, 1993). WWK had previously developed its own commercially available COO software product called COOL ONE.³¹ However, as two interviewed experts highlighted, it was crucial to have a third-party organization such as SEMATECH for spreading COO:

David Jimenez was also promoting the use of COO methodology in order to sell their commercial version [COOL ONE], but I believe it would have been very difficult for him to spread the use of COO as effectively as it was done through SEMATECH.

I agree with [the first expert] that WWK would have struggled to gain traction on COO without SEMATECH acting as an evangelical.³²

After being awarded with the contract in 1994, WWK needed to incorporate several enhancements required by the TAB into the new commercialized version, which was renamed TWO COOL (Trybula, 2006). TWO COOL was released in April 1995 and included the required features such as a database. It replaced the SEMATECH COO Revision B model, which then had over 1,200 copies in use. Daren Dance, at that time SEMATECH’s Cost Modeling Team Leader, notified existing COO users about the release of TWO COOL in a letter.³³ In June 1995, two months after the outsourcing of the SEMATECH spreadsheet model to WWK was completed, Daren Dance moved from SEMATECH to WWK as Manager, Applications Research and Development.³⁴

Standardization with SEMI

In addition to commercializing the SEMATECH model as a software package, the TAB and Users Group wanted the key components of the former SEMATECH model (e.g. equations, terminology, definitions of cost components) to become the basis for an industry standard to significantly reduce the possibility that there would be different “varieties” of COO models, as these would provide different calculated COO values that would be inconsistent with each other and therefore very difficult to compare effectively. Incomparability of different models was an issue in two kinds of situations – when competing next generations of manufacturing technology were evaluated against each other in a joint R&D setting such as at SEMATECH and when semiconductor manufacturers evaluated equipment for purchase and required information for equipment COO from suppliers. In both cases, a common understanding of definitions, inputs, and outputs is important for communication and decision making. The concern of incomparability was very real, as different COO models were being used by various IC manufacturers (although the trend was for at least those that were members of SEMATECH to standardize their models) and by different equipment suppliers. The TAB and Users Group realized they could not and did not have the authority to keep different COO software models from being used, but at least a standardized way would minimize the differences if they claimed that their software and results complied with the industry standard.³⁵ Therefore, parallel to the start of the commercialization of the COO model as a software package described in the previous subsection, standardization of the method for the calculation of COO was initiated.³⁶

In 1993, discussions on COO standardization began between SEMATECH and another semiconductor industry organization called SEMI (Lafrance and Westrate, 1993). SEMI was founded in 1970 and has been involved in the generation of industry standards since 1973.³⁷ The organization provides a forum for collaboration and standard setting, and it brings together industry experts to exchange ideas and work toward developing globally accepted standards. These are mainly technical standards (such as for production processes, testing, or wafer size), but some standards, called “equipment metrics,” concern topics such as availability, productivity, and costs. SEMI was considered the appropriate organization because of its focus on standard setting, whereas SEMATECH was focused on technical R&D (Scace, n.d.). The discussions between SEMI and SEMATECH resulted in a proposal by the TAB to form a new SEMI Standards Task Force dedicated to the development of a COO standard.³⁸ The organizational structures, such as Task Forces of SEMATECH and SEMI, allowed COO experts from different organizations to work together to develop and spread COO, as one of the people who had been part of such groups for COO for many years described:

SEMATECH provided a forum for competitors to work together, alongside with equipment and materials suppliers especially those who were members of SEMI, on common industry problems and needs on a more cost-efficient basis, while avoiding antitrust issues. SEMI was already the recognized primary semiconductor industry global standards development organization and was the natural choice to work with to develop an industry standard.³⁹

At SEMI, the “Equipment Cost of Ownership Task Force” (COO TF hereafter) was formed in 1993. Co-chairs were Daren Dance (at that point a SEMATECH employee, who also led the bidding process for the commercialization of the original SEMATECH COO model described in the previous subsection) and Dr Bob Bachrach (from Applied Materials).⁴⁰ Many of the original COO TF members and leaders were SEMATECH employees/assignees and members of the TAB and Users Group. For instance, David Bouldin who was TI’s representative on the SEMATECH Cost Modeling TAB and User Group since 1992 became one of the original COO TF members. Before that, he had been a user of other SEMI standards and therefore was familiar with the organization.⁴¹ To comply with SEMI Standards Program Regulations, the COO TF membership was not limited to US members. Also, individual companies who were not SEMATECH member companies, especially the equipment suppliers, were strongly encouraged to participate. Thus, the SEMI E35 standard development effort brought together the two major user groups of a future COO standard, namely, equipment suppliers and semiconductor manufacturers, who incurred COO when using the equipment. The participants on the COO TF, including those from the two major user groups mentioned above, were unpaid SEMI Standards Program volunteers, usually chosen and assigned by their employer. Daren Dance was the obvious choice to be one of the COO TF leaders as he worked for SEMATECH and was the key person responsible for the COO software and its development at SEMATECH at that time.⁴² The efforts of the COO TF resulted in the creation of the industry standard SEMI E35 (Sohn and Moon, 2003), which was accepted by worldwide balloting (Abramson et al., 1997).

The SEMI E35 standard was heavily influenced by the SEMATECH spreadsheet models and by the software specification for the contract with WWK mentioned in the previous subsection.⁴³ Several experts explained the relevance of the software specification for the creation of the SEMI E35 standard:

The software specification that ultimately led to TWO COOL was a direct outcome of several meetings by the SEMATECH COO Users Group. This had both SEMATECH and SEMI/SEMATECH members and predated [SEMI] E35. The specification was a very well thought out document with input from a large base already familiar with the previous SEMATECH spreadsheet releases. This specification was, in my opinion, the basis for the first drafts of [the SEMI] E35. ⁴⁴

In response, another expert who participated at the time stated,

I agree that the software specification used for the SEMATECH/WWK contract was probably a major starting point source for developing the initial draft of [the SEMI] E35 that I believe Daren [Dance] created.⁴⁵

Since the first publication of the SEMI E35 standard in 1995,⁴⁶ the software company WWK has been closely connected to SEMI and has been updating the TWO COOL software to stay compliant with this standard.

Standard revision

Revisions of the SEMI E35 standard are based on SEMI’s general standard revision regulations and procedure. These are described in two publicly available documents (SEMI, 2013a, 2013b) originating in February 1988. The regulations and procedure provide another key element for understanding the context in which the SEMI E35 standard for calculating COO could be developed. The regulations and procedure were mainly used for technical standards and also provided a basis for developing and updating the SEMI E35 standard for the calculation of COO.

A revision is triggered if the responsible Technical Committee identifies technical issues, or if the latest publication process was more than five years ago (SEMI, 2013a). The Technical Committee initiates the review by way of a standardized form, which requires information on the expected benefits, the problems of action that need addressing, the scope of work, project timetable, and intellectual property considerations, as well as which other Task Forces and committees should be kept informed (SEMI, 2013a).⁴⁷

When a revision activity is approved, it is assigned to a Task Force, which is a subgroup of a Technical Committee, and its main function is to develop standards documents. The COO TF is responsible for reviewing and updating equipment COO documents with experts from the semiconductor industry. These reviewing and updating activities also concern the consistency among related standards, such as the SEMI E35 and the SEMI E10 standards. The discussions within the COO TF focus on technical issues and lead to proposals for particular revisions of a standard. Through the participation representatives of the software firm WWK in the revision process of the SEMI E35, the firm understands how to revise their TWO COOL software to maintain compliance with the SEMI E35 standard.⁴⁸ A Task Force is expected to provide reports of the progress of the work to the Technical Committee and can also solicit ideas and feedback with an informational ballot in the form of a survey to gather general opinions. If a Task Force is not able to reach an agreement, the committee leader can delay the decision to gather other technical opinions from the expert community.⁴⁹

The meetings of Technical Committees follow a procedure that is reflected in the guidelines for the agenda. For instance, the procedure always includes a review of the program membership requirement, an antitrust reminder, and an intellectual property reminder. Furthermore, technical presentations should not include confidential material as the meetings are public forums (SEMI, 2013a). The Technical Committee votes on a Task Force’s proposals for revisions of a standard. Notably, in addition to a vote by persons, a voting system by “voting interests” exists. “Voting interests” are identified per company and are used to measure whether relevant parties (e.g. direct suppliers, indirect suppliers, end users) are sufficiently represented in the Technical Committees.

Voting on a standards ballot

A standards ballot, which is the updated version of a standard document recommended by the Technical Committee, is voted on by the entire SEMI community through a letter ballot. Voters rejecting the proposal must write at least one formal “Negative,” which is an explanation of their objection. The Negatives are crucial in the process of approving a standard – if any Negatives remain at the end of this process, approval is not forthcoming. The procedure emphasizes that every effort should be made to reach a consensus. Furthermore, specific guidelines exist with respect to classifying Negatives as not related and non-persuasive.⁵⁰ In-depth technical discussions within the COO TF focus on considering all facets and reaching agreement based on the technical arguments, which is important for the acceptance of a standard revision. This focus on reaching a consensus based on technical reasoning can also be crucial for dealing with conflicting interests, as can be seen from an example mentioned by the interviewees of a participant in COO standard development who tried to steer the SEMI E35 standard in a direction that was in their own business interest:

I agree with [another expert] that the driving force behind the development of a COO standard was not political but that is not to say that there weren’t political/business motivations behind participants in the standards development. My personal experience in that [COO] TF was that [an equipment company] tried to steer the standard in a direction that, in my opinion, would have made it fairly useless. Many years later, after observing [that same company] in other standards processes …, I came to the conclusion that standards promote competition and that companies with leading market shares feel they are not necessarily best served by standards. That is where [IC manufacturing companies] used their influence to “highly suggest compliance.”⁵¹

In the example above, a representative from the company was responsible for editing agreed-upon changes into the draft document. Yet, he did not comply and undertook changes according to his company’s interest. Ultimately, it was decided by majority vote to have editing handled by another member to solve the issue.⁵² Another expert, who had also been involved in this process, commented on the role of the established SEMI procedures in this context:

I totally agree with your [equipment company name] comment and did not think about what their rep was trying to do on their behalf as being “politically” motivated, but I can certainly see where it could be considered so from a business point of view. This is a good example where the SEMI Standards process worked effectively to prevent one company from dominating the standards development process although they did cause considerable grief and delays for us at the time.⁵³

Once a Task Force has reached an agreement, the International Standards Committee – Audits and Reviews Subcommittee determines whether all procedural requirements have been met and whether a satisfactory consensus has been reached. Subsequently, the document is approved and published (SEMI, 2013b).

In sum, the SEMI revision procedures are aimed at reaching consensus about a revision of a standard document, making sure all technical issues are acknowledged and discussed. Decisions are based on votes, but approval is not simply a matter of a majority decision. A single important objective against a proposal raised by one expert can be an overriding argument. Decision making is based on technical discussions aimed at persuasion with technical arguments, and is finalized by voting.

Discussion

The previous sections address the first research objective of this research, namely, to document the accounting story of the COO standard development stretching from the late 1980s to the mid-1990s. This section covers the remaining two objectives by reflecting on why the SEMI E35 standard emerged and what we might learn, more broadly, about standard development in management accounting.

For understanding why the SEMI E35 standard emerged, the role of COO is important. Miller and O’Leary (2007) have described how COO supports coordination of R&D among firms because it helps to evaluate candidate technologies. These technologies are compared in terms of COO and against semiconductor industry expectations for ongoing cost reduction to see which technology is worth pursuing in the R&D programs of the various organizations (Hazelton et al., 2008). The description of SEMI E35 and SEMI E10 demonstrated the complexity of the calculations and the necessary degree of detail in definitions. As the semiconductor industry cooperated on R&D and COO was important for technology choices and investment decisions, the merits of having a “common language” codified by a standard were apparent.

Enabling conditions for the existence of this standard

However, the role of COO as mentioned above does not yet explain why a standard would actually exist. In this section, we discuss how standard development was indirectly influenced by the existence of technical standards in three ways.

First, network organizations existed for joint research and technical standard setting, which then also provided a platform for development of the COO standard. Existing network organizations such as SEMATECH (Carayannis and Alexander, 2004) provided an umbrella for semiconductor industry players throughout the supply chain to work together, even as competitors. SEMI has been involved in the generation of technical semiconductor industry standards since 1973 and provided a forum for collaboration and technical standard setting. Thus, companies had experience with joint research conducted and promoted through these network organizations.⁵⁴ Having a third-party organization such as SEMATECH was also considered instrumental for spreading the use of COO. Several people working for SEMATECH member companies became assignees and worked at SEMATECH for a period of time, thus contributing and distributing COO knowledge within that network. A very relevant example for this is the Intel assignee Dean Toombs, who brought the concept of COO to SEMATECH. Furthermore, these network organizations provided specific institutional arrangements for the cooperation of experts from various companies and other organizations during a certain period on a particular topic, such as the TABs and the User Groups at SEMATECH, and the Technical Committees and the Task Forces at SEMI. As antitrust issues were avoided in these institutions, competitors could cooperate on solving common semiconductor industry problems such as technology evaluation or the COO standard development.

Second, connections between individuals involved in COO and their mobility played an important role. Many of these individuals already knew each other across boundaries of companies because of prior collaboration on technical topics, and they moved between organizations circulating COO knowledge. Informal contacts existed between COO experts in different organizations, such as between the groups at TI and Intel that were working on their own COO models. Also, experts who became involved in COO development did not start with an entirely new network, as they could build on their existing connections and expertise in technical areas, such as yield modeling, as input for cooperation. The example of the cooperation between Daren Dance and Charles Stapper mentioned above is illustrative. They were already experts on yield modeling and subsequently cooperated in developing COO models. People have been working together over a considerable period, and several of them are still doing so today. Preexisting contacts also included the mentioned connections between David Jimenez and SEMI/SEMATECH prior to his firm WWK being selected for commercialization of the SEMATECH model. Individuals moved between organizations and thereby disseminated COO knowledge such as Dean Toombs, who moved between TI, Intel, and SEMATECH and promoted COO modeling; Daren Dance, who went from SEMATECH to the software firm WWK; and David Bouldin, who continued his work as an assigned expert when the COO migrated from the SEMATECH TAB and the User Group to the SEMI Technical Committee and the COO TF. In another “move,” SEMATECH and many of its members now use the software that WWK had developed as a further refinement of SEMATECH’s spreadsheet models. The move from SEMATECH to WWK was relevant for the creation of the SEMI E35 standard as the software specification in the contract served as the starting point for the initial SEMI E35 standard.

Third, apart from relying on the networks of experts, the COO development could build on the experience that was documented in SEMI’s procedures for joint industry standard development. These procedures, which codified SEMI’s experience, have been used to generate a great number of standards, mainly on technical topics. One of these technical standards is the SEMI E10 that defines equipment states. It has existed since 1986 and has been closely linked to early COO models, such as the one published by Ross Carnes in 1991, and the resulting SEMI E35 COO standard (Carnes and Su, 1991). The procedures were helpful when the SEMI E35 standard for the calculation of COO needed to be developed and updated as participants did not need to start from scratch. These procedures are focused on creating consensus through technical discussions and on making sure that all expert inputs are given weight and treated in a transparent way. Conflicts of interests during standard development are sometimes complicated, and appropriate processes prevent a company from dominating the development process.

The context around technical standards may enable the development of standards for management accounting

In this last part of the discussion section, we want to synthesize our interpretations of the spill-overs that can be obtained more broadly regarding standard development in management accounting. As a new insight, this study suggests that the context existing around technical standards may also influence the development of standards for management accounting, and as far as we are aware, this influence has hardly been described in prior literature. Network organizations for joint research and technical standard setting, connections and mobility of people for technical topics, and existing procedures and codified experiences from technical standard development may also be factors to consider in other industries where management accounting standards could be relevant. These specific factors may help to better understand why standards for management accounting information that is exchanged beyond the boundaries of a single firm may be difficult to establish. For example, our interpretation could suggest that without the presence of network organizations that exist for technical reasons, the mere wish to develop management accounting standards may not be enough to enable organizations and people in many other industries to successfully cooperate for that purpose alone. Our study suggests that it is the rare combination of such enabling circumstances – in particular the organization that handles the standards – and the real requirements for the standards from the user and producer groups represented by the organization that may be needed for the existence of management accounting standards.

The cooperation for standard development between IC manufacturers and equipment suppliers in network organizations such as SEMI and SEMATECH provides an interesting contrast to the example of the “Blue Book” mentioned in the introduction section (Jack, 2015). The “Blue Book” failed partially because of the “supply-led” nature of the standardization effort; farmers and graziers were left out of the development process. Ultimately, the offer was meeting a need that was recognized by advisors but not by the future users of that standard (e.g. the farmers). In the semiconductor sector, COO was a mediating instrument and the development of the SEMI E35 standard was driven by the needs of users (Miller and O’Leary, 2007). The users of the COO standard were not a homogeneous group, but they included equipment manufacturers and also IC companies, who incurred COO when using the equipment. The SEMI E35 standard development was stimulated by several producers of this accounting innovation (i.e. SEMATECH, SEMI, the software firm WWK). Users became producers of accounting standards because people moved between organizations and represented both users (as employees of firms) and producers (when participating in SEMATECH and SEMI COO projects). Thus, the SEMI E35 standard development was also a function of the needs of the users.

Conclusion

The SEMI standard for the calculation of COO seems to be a rare example of a voluntary, public, and widely used standard for interorganizational management accounting (Geissdörfer et al., 2009). This research contributes to accounting history research with new data and interpretations along its three objectives summarized below.

The principal objective of this article was to document an interesting management accounting episode of standard development. We recount the history of the development of the SEMI E35 standard for the calculation of COO in the semiconductor industry from the late 1980s to the mid-1990s, which to our knowledge has not yet been documented in a coherent way.

The second objective was to discuss factors that may explain why this development occurred. COO modeling efforts became more common across several organizations. Exchangeability of such information made standards increasingly necessary. Industry organizations such as SEMATECH and SEMI provided structures for people to work together on standardization. In addition, networks of experts existed, underlying technical standards such as the SEMI E10 were in place, and practices for standard development were available.

Our third objective was to reflect on what we might learn more broadly from this COO standard about standard development in management accounting. All of the conditions mentioned above under our second objective existed for other reasons, particularly R&D cooperation and development of technical standards. Therefore, the step to also develop and use a standard for COO calculations became a small one within this enabling context of organizations and practices for standard development. However, management accounting standards that are widely used in an industry are very rare. In other industries – even if there would be real requirements for such standards – establishing standards for management accounting may be too difficult because of the absence of a related context. This study suggests that network organizations for cooperation and practices that can be leveraged for management accounting standard creation are an integral part of such a context.

We combined public data from different sources with private data, such as interviews with several experts, which enabled triangulating of the discoveries. Future research could further analyze this management accounting story employing an explicit theory-based approach. The present study may be a stepping stone for future research to look in more detail at these, what Yates (2005) calls “structuring factors.” The Yate’s study investigated the adoption of information technology (starting with the punched-card tabulator) in the US life insurance industry in the period from around 1890 until around 1980.⁵⁵ Relying on structuration theory, she investigates how innovations with information technology were the product of complex interactions between structuring factors such as the needs of companies as technology users, individual managers, and technical experts as advocates for use of new technologies working inside firms, external advocates such as professional and trade associations, regulators, and vendors of the new technology. The case of COO standardization in the semiconductor industry could be an interesting one to analyze through the lenses of structuration theory. Further investigation along that framework might focus on “needs” of technology users for COO to evaluate technologies, “individual advocates” such as Dean Toombs, and “external advocates” such as SEMATECH. Future research could also look at other industries and investigate to what extent the conditions we identified may explain the absence or presence of interorganizational standards for management accounting in other industries. Industries to investigate could be those close to the semiconductor industry, such as photovoltaic (Brown et al., 2010; Raithel et al., 2013), or more distant industries such as the food equipment industry, which has announced a COO initiative (White, 2006).

Footnotes

Appendix 1 Acknowledgements

The authors thank the industry experts we could interview, in particular David Jimenez, David Bouldin, Stephan Raithel, Walt Trybula, Markus Lentz, Scotten W. Jones, Diederik de Bruin, Boudewijn Sluijk, Andrea Wüest, Peter Wagner, Wolfgang Herbst, Arun Ramakrishnan and James Amano. We also thank the editor and reviewers for their extremely helpful comments and suggestions.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

Notes

ORCID iD

Maximilian Sandholzer

References

Abramson

Encarnacao

Reid

Schmoch

, National Academy of Engineering, and Fraunhofer Institute for Systems and Innovation Research (1997) Technology Transfer Systems in the United States and Germany: Lessons and Perspectives. Washington, DC: National Academies Press.

Bisman

(2012) Surveying the landscape: The first 15 years of Accounting History as an international journal. Accounting History 17(1): 5–34.

Brown

Linden

(2009) Crisis 1: Loss of competitive advantage. In: Chips and Change. How Crisis Reshapes the Semiconductor Industry. Cambridge, MA: MIT Press, pp. 15–38.

Brown

Jimenez

Williams

(2010) Examining cost of ownership for front- and back-side metallization of crystalline-silicon solar cells. Photovoltaics International 8: 1–7.

Caglio

Ditillo

(2008) A review and discussion of management control in inter-firm relationships: Achievements and future directions. Accounting, Organizations and Society 33(7): 865–898.

Camfferman

(2012) A contract-law perspective on legal cases in financial reporting: The Netherlands, 1880–1970. Accounting History 17(2): 141–173.

Camfferman

Zeff

(2015) Aiming for Global Accounting Standards: The International Accounting Standards Board, 2001-2011. Oxford: Oxford University Press.

Carayannis

Alexander

(2004) Strategy, structure, and performance issues of precompetitive R&D consortia: Insights and lessons learned from SEMATECH. IEEE Transactions on Engineering Management 51(2): 226–232.

Carnes

(1991) Long term cost of ownership: Beyond purchase price (chip manufacture). In: IEEE/SEMI international semiconductor manufacturing science symposium, 1991 (ISMSS 1991), Burlingame, CA, 20–22 May. New York: IEEE.

10.

Chandar

Collier

Miranti

(2012) Graph standardization and management accounting at AT&T during the 1920s. Accounting History 17(1): 35–62.

11.

Collier

(2012) A contextual analysis of the development and diffusion of depreciation accounting at the Bell System, 1910–1937. Accounting History Review 22(1): 23–45.

12.

Coronella

Lombrano

Zanin

(2013) State accounting innovations in pre-unification Italy. Accounting History Review 23(1): 1–21.

13.

Cortese

(2011) Standardizing oil and gas accounting in the US in the 1970s: Insights from the perspective of regulatory capture. Accounting History 16(4): 403–421.

14.

Dance

DiFloria

Jimenez

(1996) Modeling the cost of ownership of assembly and inspection. IEEE Transactions on Components, Packaging and Manufacturing Technology: Part C: Manufacturing 19(1): 57–60.

15.

Dekker

(2003) Value chain analysis in interfirm relationships: A field study. Management Accounting Research 14(1): 1–23.

16.

Dekker

Van Goor

(2000) Supply chain management and management accounting: A case study of activity-based costing. International Journal of Logistics Research and Applications 3(1): 41–52.

17.

Ding

McKinstry

(2013) Paper trails: The development of management accounting at Alex. Cowan & Sons Ltd, Penicuik, 1779–1965. Accounting History 18(1): 99–119.

18.

Ellram

Siferd

(1993) Purchasing: The cornerstone of the total cost of ownership concept. Journal of Business Logistics 14(1): 163–184.

19.

Geissdörfer

Gleich

Wald

(2009) Potentials of standardizing life cycle cost analysis models for strategic cost management. Zeitschrift für Betriebswirtschaft 79(6): 693–715.

20.

Granlund

(2011) Extending AIS research to management accounting and control issues: A research note. International Journal of Accounting Information Systems 12(1): 3–19.

21.

Hazelton

Wüest

Hughes

Lercel

(2008) Predicting lithography costs: Guidance for ≤ 32 nm patterning solutions. In: Proceedings of the SPIE 7028, Photomask and Next-Generation Lithography Mask Technology XV, 70283N (ed. Horiuchi

), Yokohama, Japan, 19 May, The International Society for Optical Engineering, p. 70283N.

22.

Heilala

Helin

Montonen

(2006) Total cost of ownership analysis for modular final assembly systems. International Journal of Production Research 44(18–19): 3967–3988.

23.

Hergert

Morris

(1989) Accounting data for value chain analysis. Strategic Management Journal 10(2): 175–188.

24.

Irwin

(1996) The U.S.–Japan Semiconductor trade conflict. In: Krueger

(ed.) The Political Economy of Trade Protection. Chicago, IL: The University of Chicago Press, pp. 5–14.

25.

Jack

(2015) Future making in farm management accounting: The Australian “blue book.” Accounting History 20(2): 158–182.

26.

Kajüter

Kulmala

(2005) Open-book accounting in networks. Management Accounting Research 16(2): 179–204.

27.

Lafrance

Westrate

(1993) Cost of ownership: The supplier’s view. Solid State Technology 36(7): 33–37.

28.

LaLonde

Pohlen

(1996) Issues in supply chain costing. International Journal of Logistics Management 7(1): 1–12.

29.

Levant

Nikitin

(2012) Can cost and financial accounting be fully re-integrated? The role of the French state in the separation of accounting systems and the failed attempt of the système croisé to re-integrate them. Accounting History 17(3–4): 437–461.

30.

Levant

Zimnovitch

(2013) Contemporary evolutions in costing methods: Understanding these trends through the use of equivalence methods in France. Accounting History 18(1): 51–75.

31.

McIvor

(2001) Lean supply: The design and cost reduction dimensions. European Journal of Purchasing & Supply Management 7(4): 227–242.

32.

Mckeen

Smith

(2010) Developments in practice XXXVII: Total cost of ownership. Communications of the Association for Information Systems 27(32): 627–636.

33.

Mieritz

Kirwin

(2005) Defining Gartner Total Cost of Ownership. Stamford, CT: Gartner, Inc.

34.

Miller

O’Leary

(2007) Mediating instruments and making markets: Capital budgeting, science and the economy. Accounting, Organizations and Society 32(7–8): 701–734.

35.

Miller

Moll

O’Leary

(2012) Managing Inter-Firm Interdependencies in R & D Investment: Insights from the Semiconductor Industry. London: Chartered Institute of Management Accountants.

36.

Oechsner

Pfeffer

Pfitzner

Binder

Müller

Vonderstrass

(2002) From overall equipment efficiency (OEE) to overall Fab effectiveness (OFE). Materials Science in Semiconductor Processing 5(4–5): 333–339.

37.

Ragona

(2002) Cost of ownership (COO) for optoelectronic manufacturing equipment. In: 2002 microsystems conference, Victor, NY. Available at: https://pdfs.semanticscholar.org/f543/552c66f67fb2b468ad1bc7d6e9df74c68c5a.pdf

38.

Raithel

Amano

Bouldin

Wessendorf

Buenting

(2013) Total Cost of Ownership in PV Manufacturing Guide. SEMI and VDMA.

39.

Richardson

(2011) Regulatory competition in accounting: A history of the Accounting Standards Authority of Canada. Accounting History Review 21(1): 95–114.

40.

Scace

(n.d.) Forty years of semi standards! Available at: http://www.semi.org/en/Standards/P043719 (accessed 7 March 2018).

41.

Schäffer

Binder

(2008) “Controlling” as an academic discipline: The development of management accounting and management control research in German-speaking countries between 1970 and 2003. Accounting History 13(1): 33–74.

42.

Schulze

Seuring

Ewering

(2012) Applying activity-based costing in a supply chain environment. International Journal of Production Economics 135(2): 716–725.

43.

Seal

Cullen

Dunlop

Berry

Ahmed

(1999) Enacting a European supply chain: A case study on the role of management accounting. Management Accounting Research 10(3): 303–322.

44.

Seidel

(2007) EUV lithography cost of ownership considerations. In: International EUVL symposium, Sapporo, Japan, 29 October. Available at: http://www.sematech.org/meetings/archives/litho/8059/pres/CO-01-Seidel.pdf

45.

SEMI (2012a) Guide to Calculate Cost of Ownership (COO) Metrics for Semiconductor Manufacturing Equipment E35-0312. San Jose, CA: SEMI.

46.

SEMI (2012b) Specification for Definition and Measurement of Equipment Reliability, Availability, and Maintainability (RAM) and Utilization E35-0312. San Jose, CA: SEMI.

47.

SEMI (2013a) Procedure Guide Detailing SEMI Standards Processes and Practices. San Jose, CA: SEMI.

48.

SEMI (2013b) Regulations Governing SEMI Standards Program. San Jose, CA: SEMI.

49.

Shoven

Topper

(1992) The cost of capital in Canada, the United States, and Japan. In: Shoven

Whalley

(eds) Canada-U.S. Tax Comparisons. Chicago, IL: The University of Chicago Press, pp. 217–236.

50.

Silverman

(2005) Extreme ultraviolet lithography: Overview and development status. Journal of Micro/Nanolithography, MEMS, and MOEMS 4(1): 11006.

51.

Sohn

Moon

(2003) Cost of ownership model for inspection of multiple quality attributes. IEEE Transactions on Semiconductor Manufacturing 16(3): 565–571.

52.

Spencer

Seidel

(1995) National technology roadmaps: The U.S. semiconductor experience. In: Proceedings of 4th International Conference on Solid-State and IC Technology, IEEE, pp. 211–220. Available at: http://ieeexplore.ieee.org/document/500069/

53.

Trybula

(2006) Cost of ownership – Projecting the future. Microelectronic Engineering 83(4–9): 614–618.

54.

White

(2006) Determining total cost of ownership. Foodservice Equipment & Supplies 59(2): 28–32.

55.

Yates

(2005) Structuring the Information Age: Life Insurance in Technology in the Twentieth Century. Baltimore, MD: Johns Hopkins University Press.