An Updated Inquiry into the Study of Corporate Codes of Ethics: 2005–2016

This paper presents a review of 100 empirical papers studying corporate codes of ethics (CCEs) in business organizations from the time period mid-2005 until mid-2016, following approximately an 11-year time period after the previous review of the literature. The reviewed papers are broadly categorized as content-oriented, output-oriented, or transformation-oriented. The review sheds light on empirical focus, context, questions addressed, methods, findings and theory. The findings are discussed in terms of the three categories as well as the aggregate, stock of empirical CCE studies in comparison with previous reviews, answering the question “where are we now?” Content and output studies still stand for the majority of the studies, whereas the transformation studies are fewer. Within these areas, two new trends are found to have emerged: discursive analyses and a focus on labor conditions. The review finds that (a) the content of CCEs is still predominantly self-defensive, (b) that CCEs are insufficient in themselves in terms of protecting workers’ rights, (c) that CCEs are likely to encounter tensions when implemented across national and organizational boundaries, and (d) that while perception of CCEs is generally positive, CCEs may lead to both positive and negative outcomes. Based on these findings, potential areas for further exploration in the area of CCE research are suggested.


Introduction
Most modern corporations have a corporate code of ethics. Sharbatoghlie et al. (2013) report that 95% of both Fortune US 100 and Fortune Global 100 companies have a code of ethics. Pitt and Grsokaufmanis (1990) note that "[a]dopting a code of conduct is tantamount to a commitment to engage in corporate self-regulation" (p. 1) and provide one of the earlier definitions of corporate codes, which they define as "written statement of ethics, law, or policy (or some combination thereof), delineating the obligations of one or more classes of corporate employees." More recently, Stevens (1996, p. 72) defines ethical codes as "written documents which presume to state the major philosophical principles of an organization." Kaptein and Wempe (1998) state that "a code is a suitable instrument to increase the moral resistance of an organization" (p. 854). Schwartz (2001, p. 248) defines a corporate code of ethics as "a written, distinct and formal document which consists of moral standards used to guide employee or corporate behavior." Most recently, a succinct version of the Kaptein and Schwartz (2008) 1 definition of a business code is provided by Kaptein (2011), in that, a business code of ethics is "a set [of] prescriptions developed by a company to guide the behavior of managers and employees." Taking these definitions together, and for the purposes of our study, we define Corporate Code of Ethics (CCE 2 ) as written and formal documents intended to increase moral resistance 1 Kaptein and Schwartz (2008, p. 113) define a business code as "a distinct and formal document containing a set of prescriptions developed by and for a company to guide present and future behavior on multiple issues of at least its managers and employees toward one another, the company, external stakeholders and/or society in general." 2 CCE is consequently used throughout the paper even if the papers in the review use concepts like Code of Conducts, Business Code of Ethics or similar.

3
in the organization and to guide corporate, employee and other stakeholders' behavior. To be clear, our definition of CCEs refers exclusively to company codes and excludes codes from other levels of business, e.g., professional codes and non-government organization (NGO) codes. 3 We also exclude company mission statements because " [m]ission statements declare what the corporation intends to accomplish, while ethical codes address the values embraced by the corporation" (Stevens 1996, p. 72).
Although the recent decade has seen an increase in the conspicuousness of CCEs as parts of corporate governance, these are not a new phenomenon. CCEs have been used in business and have been a subject of interest in business studies for at least a hundred years (Graves 1924;Taeusch 1932). Companies install CCEs for different reasons, including as a means to prevent and hinder unethical behavior, even though the code "is not a cure-all, and it possesses no magic powers by which it can change moral darkness into light" (Graves 1924, p. 59). Other reasons for code adoption include meeting legal requirements for stock exchange listing, guiding employees, communicating principles to stakeholders, and protecting and enhancing reputation (Bondy et al. 2004) as well as imitative behavior (Holder-Webb and Cohen 2012). Although CCEs are not a new phenomenon, they are increasingly in use and have become a common phenomenon in the modern corporation.
The beginning of the mass upsurge of CCEs is possibly attributable to legislation such as the Sarbanes-Oxley Act in 2002 followed by amendments in the U.S. Sentencing Guidelines for Organizations in 2004, outlining guidelines for effective compliance and ethics programs. CCEs are today an integral and expected management tool in large corporations. Understanding what CCEs are, as well as their effects within and beyond the organization, has been of interest to many researchers using different empirical settings, methodologies, and theoretical approaches. So what have we really learned from the stock of research on CCEs so far? In the mid-90 s, Stevens (1994) conducted a review of the literature while asking the rhetorical question "do we know what we need to know about CCEs?" followed by "where do we go from here?" She concluded that most studies were content analyses and that there was a lack of studies focusing on how codes are communicated in the organization as well as a lack of evidence relating to the "golden question" of whether codes are effective, or not. She argues for studies oriented more toward a rhetorical analysis of the messages in the CCEs. More specifically, Stevens called for studies seeking answers to the following questions: "do they work?…are the codes communicated in meaningful ways? Are employees aware of their organization's ethical code and accepting of its guiding principles?" (1994, p. 68).
Taking heed of the focus of the literature review by Stevens (1994) and covering the years from 1994 to mid-2005, Helin and Sandström (2007) review empirical studies on CCEs and make several interesting conclusions. Categorizing the papers within three broad areas, defined as "content oriented (what is in the actual codes), output oriented (what effects they have), or transformation-oriented (how the codes are coming into practice or not in the organization)" (Helin and Sandström 2007, p. 254), the review gives an overview of the empirical studies in the field. The review concludes that a majority of the studies deal with code content and that these studies show "CCEs in different countries are more or less the same, but that there also are some country-specific issues" (2007, p. 261). The authors reflect that the focus of the content studies is exploratory rather than explanatory. Regarding the output-oriented studies, the authors find "an evident lack of insights into how CCEs influence behavior in organizations" and " [B]ehavior related to CCEs seems to be a question of perception, not action" (2007, p. 262). Relating to the transformation-oriented papers, the authors find a gap in how codes and behavior are transformed, and conclude " [W]hat is still lacking is how this process of contextualization is carried out. What kinds of problems arise? Which actors translate CCEs? How is their behavior altered? What kinds of transformations take place?" (2007, p. 262). Furthermore, the review suggests that research studies have employed a wide range of research methods but with a weak use of explicit theory. Helin and Sandström thus highlight a gap in the literature regarding the process and transformation of CCEs, that is questions pertaining to how codes are communicated, how are they translated into daily action, how they (do or do not) influence behavior? Kaptein and Schwartz (2008) and Stevens (2008) review empirical studies examining code effectiveness, i.e., output-oriented studies, and explore why the aggregated results of these studies are inconclusive. Kaptein and Schwartz (2008) argue that conflicting results regarding code effectiveness are due to "varying definitions of key terms, deficiencies in data and methodologies used, as well as a lack of theory" (p. 111). Suggesting an integrated research model for the study of the effectiveness of business codes, Kaptein and Schwartz emphasize the need to understand how processes inside and outside the firm are carried out as well as how they relate to one another, and propose the fourteen factors (see Fig. 1, 2008, p. 118) of their integrated research model be taken into consideration in future studies of the effectiveness of business codes. The authors also propose that researchers distinguish between actual and perceived code effectiveness. Based on their analysis of the literature, Kaptein and Schwartz (2008) believe that future research of the effectiveness of codes is best performed by studying "multiple companies in which the factors […] are longitudinally measured before and after the introduction of the business code" (2008, p. 122). As such, Kaptein and Schwartz (2008) provide guidance, direction, and goals for future research of the effectiveness of codes, i.e., output-oriented studies. Stevens (2008) adds to the research by asking when codes are effective and when they are not. Consistent with her Stevens (1994) paper, she emphasizes the importance of communication in the processes of implementing CCEs. Her review looks at four studies demonstrating the ineffectiveness of CCE and eight studies indicating effectiveness, i.e., 12 output-oriented studies. Stevens finds that CCEs implemented as a control system and/or imposed on an organization from the top down were ineffective, while effective CCEs were dependent upon integrated and embedded cultural values and effective communication., Stevens (2008) states that codes are effective, and calls for ethical research to focus on creating and implementing codes that articulate the values of the corporation, influence behavior in accordance with the values throughout the corporation, and have strong communication channels.
With over a decade having passed since the last publication of a code of ethics literature review, this paper aims to review the current status of empirical research on CCEs, covering the period from mid-2005 to mid-2016. The purpose of this study is to provide an updated answer to the question identified by Stevens (1994) and approached by Helin and Sandström (2007), by asking a further decade later, "where are we now?" To answer this question, this review of the literature concentrates on the areas of empirical focus, contextual settings, questions addressed, theories and methodologies used, and the findings of the studies. A particular focus is directed towards learning whether the gap related to how codes (do and do not) work in practice, and when codes are effective or not. Helin and Sandström's (2007) three original categories are used for the overarching categorization regarding questions being addressed in the studies (content, output and transformation) and are defined as: Content oriented (what is in the actual codes?), Output oriented (what effects they have?), and Transformation-oriented (how the codes are coming into practice or not in the organization?). The three categories aid in answering parts of the overall question of 'where are we now', such that the content-oriented studies can tell us about the current content and focus of codes within business organizations, the output-oriented studies can shed further light on whether and when codes are effective, and the transformation-oriented studies can explore processes and consequences of codes in practice.
To deliver this purpose, the paper is structured as follows. In the next section, the research methodology behind the paper is presented. This is followed by a review of the studies published between mid-2005 to mid-2016, by overarching category (main questions addressed) and respective subcategories (empirically derived). In the discussion section an overview of the findings from the review is presented and contrasted to the findings of earlier reviews. To conclude, the purpose of the study is answered, and directions for future code studies are outlined.

Methodology
This review consists of papers published in the English language in international peer-reviewed journals during the period of mid-2005 to mid-2016. For the search of articles, we used the Web of Science search engine. Corporate codes of ethics have also been referred to as codes of conduct, business codes of ethics, and in a few instances codes of practice and integrity codes (see Schwartz 2001;Kaptein 2004). The precursor to the CCE, which became pervasive in the U.S. during the 1980s, were referred to as corporate credos or creeds (Benson 1989). Consistent with the general terminology used for CCEs, and with the search criteria of Helin and Sandström (2007), the keywords used in the search were "corporate" or "business" together with "code of ethics" or "code of conduct" (corporate AND code of ethics OR corporate AND code of conduct) and (business AND code of ethics OR business AND code of conduct). These searches resulted in 498 and 929 scholarly hits, respectively. When accumulated, and after removing duplicates, the complete list contained references to 701 papers. The abstracts of these 701 papers were compiled into a single document, cataloged to a corresponding Excel worksheet, and subsequently reviewed using the same criteria as Helin and Sandström (2007), i.e., Only papers addressing CCEs in business organizations and containing empirical material were selected for inclusion in the review.
The term business organizations excludes papers dealing with non-governmental organizations (NGO), or professional groups, such as accountants, nurses, financial analysts, HR managers etc., or associations (see for example Harkrider et al. 2012;Salierno 2007;Voskuijl and Evers 2007). Empirical material means that only studies utilizing material from company official statements or reports, company employees or managers, or through ethnographic data collection are included. As such, empirical material includes interview, survey, observations, collection of data from webpages, or other types of document studies. Studies of a conceptual nature as well as reviews were excluded (see for example Pedraza and Gajiwala 2012;Seidl 2007). Studies from surveys or experiments utilizing undergraduate student participants as respondents were excluded (see for example Davidson and Stevens 2013;Gino and Margolis 2011;Umphress et al. 2009), but studies involving MBA students were included if the MBA students were also experienced business managers (see for example Hoivik 2007). The argument for excluding studies with student participants is the lack of participants' professional experience. In other excluded papers, the CCEs are mentioned but not studied explicitly (see for example MacLean et al. 2015). These papers have not been included in the review.
When reviewing the initial 701 abstracts, all three researchers read all the abstracts, and, based on the criteria mentioned above, determined independently which papers to include by indicating in an individual copy of the Excel worksheet, a "Yes" for inclusion, a "No" for exclusion, or a "Maybe" for further elaboration. Each of the researcher's lists were then compiled into one worksheet to allow for comparison. In the few instances where the three researchers did not indicate a unanimous decision about whether or not to include a paper, the following two-step process occurred. First, the consensus applied when at least two researchers were in agreement (two "Yes" or two "No" decisions). Second, papers where there was no agreement between the researchers' decisions, or where the researchers had indicated a "Maybe" decision, were discussed further among the three researchers to come to a determination. Most papers determined to be "Maybe" were included at this point.
The above procedure reduced the list down to 107 papers. The three researchers divided these papers among themselves and read 35 or 36 papers each. During this close examination of the full paper, seven papers were rejected because they did not meet the criteria. Detailed notes were made on empirical focus, context, questions addressed, theory, methods, and findings, and a summary of the paper. The researchers also preliminarily categorized each paper using the three categories of: "content," "output," or "transformation." Papers that address more than one of these categories were labeled as such, e.g., "content and output." For example, Donker et al. (2008) examine the content of CCEs for the inclusion of positive normative values and test for correlations between a measure of the level of normative value inclusion in a company's CCE and the company's economic performance. Content is the most prominent category of the overlapping categories (content and output) for the Donker et al. study, hence, the paper is reviewed under the content category, and labeled as a content/output-oriented study. Similarly, the content category contains papers labeled content/transformation-oriented studies, the output category contains papers labeled output/content-oriented studies and output/transformation-oriented studies, and the transformation category contains papers labeled transformation/outputoriented studies. Tables listing the papers reviewed within each category can be found in the next section, Review of Studies, under the respective subsections: 3.1 Content-Oriented Studies; 3.2 Output-Oriented Studies; and 3.3 Transformation-Oriented Studies. Within each table, the paper's author and publication date are provided along with subcategory classification, theory, method, data collection date, context and level of analysis, antecedents, and consequences.

Review of Studies
This section of the paper reviews empirical papers studying CCEs from the time period mid-2005 until mid-2016. Our review begins with the content-oriented studies, followed by studies categorized as output-oriented and then transformation-oriented.

Content-Oriented Studies
Content-oriented studies are still prevalent in the CCE literature and constitute close to 50% of the papers in our review of one hundred papers. These papers are subcategorized based on the primary research question addressed consistent with the initial categorization of papers into the three areas of study. This subcategorization allows for more clarity in the analysis of similarities and differences, but the review also examined the papers across other variables, e.g., theory, method, data collection date, etc. While there is obviously some overlap for studies across subcategory, the subcategories are: country specific, country comparison, or global; changes across time; industry specific; supplier codes; ethics quality; language (symbolic aspects); and CCE support structures.
Within the reviewed papers classified as content-oriented, about one-third focus on within or across country similarities and differences in code content or code support structures. Content-oriented studies also continue to examine specific industries: advertising, banking, and electronics during the current review period. New areas of focus during the last decade include examination of code content for changes across time and examination of the content in MNC supplier codes. Additionally, two new areas that hold promise for further exploration, especially the latter, are the examination of the ethics quality contained in the codes and the symbolic aspects of code content.
Prior reviews of code of ethics research by Stevens (1994) and Helin and Sandström (2007) report that, in general, results from content-oriented research suggest the primary roles of CCEs are self-defense (protection) and windowdressing, and that codes tend to be inward-looking with little regard for external stakeholders. Based on the evidence reviewed some two decades later the content of CCEs, in general, appear to be heavily influenced by corporate legal departments and are still predominantly inward looking. Table 1 provides a list of the papers reviewed under the content-oriented category including author(s) and publication date, subcategory, theory, method, data collection date, context and level of analysis, antecedents, and consequences. The table is followed by an overall summary of the papers reviewed under this category, and a review of the individual papers by subcategory.

Summary of Content-Oriented Studies
Most content-oriented studies still rely on responses to survey questionnaires or the collection of codes from corporations. However, during the last decade, there has been an increasing preponderance to collect codes directly from company websites rather than requesting copies from company personnel, which perhaps reflects greater transparency by corporations. In general, the content-oriented studies lack, or contain relatively weak, theoretical underpinnings and continue to be more exploratory than explanatory. Most studies are based on suggestions and empirical findings from prior literature, although for papers published in 2007 and beyond we see the emergence of explicit theory to motivate and direct the study (e.g., Halff 2010;Holder-Webb and Cohen 2012;Long and Driscoll 2008). This emergence of the utilization of explicit theory within content papers coincides with the expansion in methodology from a tendency to rely on frequency of mention as the measure for analysis of code content to more sophisticated measures such as lexical software analysis (Béthoux et al. 2007), centering resonance analysis (Canary and Jennings 2008), and critical discourse analysis (Winkler 2011).
Unlike the output-oriented studies that have distinct antecedents and consequences, the nature of content-oriented studies do not lend themselves to a direct examination of antecedents and consequences. The one study categorized as content-oriented that involves antecedents and consequences is Canary and Jennings (2008) which examines differences and similarities in CCE content for U.S. companies pre-and post-SOX. Similarly, there is only one study categorized both as content-and transformation-oriented with antecedents and consequences. Forsyth and O'Boyle's (2011) find the greater the degree of relativism expressed by a country's residents the less likely are the businesses within that country to develop extensive CCEs, but no support for a relationship between degrees of idealism and CCE extensiveness.
In comparison, studies classified as both content-oriented and output-oriented are expected to involve antecedents and consequences due to the nature of their output-oriented aspects. Of the six papers within this category, Donker et al. (2008) find that the level of ethics quality contained in a company's code is positively related to that company's economic performance. Using CCE content quality as the consequence, Garegnani et al. (2015) find CCE content quality is positively related to firm size but is negatively related to the firm's degree of internationalization. The other four content/output-oriented studies examine the association between governance structures and CCE implementation and embeddedness (Garcia-Sanchez et al. 2013, 2015aRodriguez-Dominguez et al. 2009). While CCE implementation or embeddedness is denoted as the dependent variable in the regression analyses for these four studies, it is unclear which variable is the antecedent and which is the consequence. Specifically, it is unclear if a company's governance structure causes a company to adopt a code, or if a company having a code has an effect on that company's governance structure.
Studies that report on changes across time pre-and post-SOX for U.S. and Canadian companies suggest that post-SOX there is an increased emphasis on legal and compliance issues in code content (Canary and Jennings 2008;Singh 2006). Other studies also show that there is a general trend in code content of corporations falling under the Anglo governance model, post-SOX, to be prescriptive in nature and include an increased emphasis on legal and regulatory aspects within code content (e.g., Singh et al. 2011). Consistent with this increased emphasis, the codes of Anglo corporations continue to be focused on protection of the firm and exhibit a greater concern about actions against the firm than actions by the corporation (e.g., Canary and Jennings 2008;Singh 2006;Singh et al. 2011).
Studies of MNC supplier codes also consistently find that codes are protectionist devices that attempt to pass responsibility down to the supplier and are little more than window dressing in an attempt to maintain profits (e.g., Preuss 2009;. Providing support for the legal influences on code content, Oehmen et al. (2010) document action research involving the creation and implementation of codes of conduct for a supplier of a mid-sized Swiss company and report that topics in the initial customized code were drastically reduced due to reservations by the company's legal department. It should also be noted that studies examining CCE content of TNCs consistently find that codes are generally not adaptive to local laws, values and customs when traveling across countries (e.g., Amaeshi and Amao 2009;Garegnani et al. 2015;Halff 2010;Svensson and Wood 2007). Perhaps the legal influence on code content could explain this lack of flexibility, and as such, may be an area worthy of further examination.
A fairly recent area within the content papers is an examination of the language, i.e., symbolic aspects, contained in the code. In other words, this line of research studies the use of language in the code rather than just the words. Overall, the studies find that, consistent with other studies utilizing less-sophisticated research methodologies, CCEs focus on firm protectionism and legitimacy with little focus on corporate responsibility (Béthoux et al. 2007;Long and Driscoll   . However, findings from studies examining the symbolic aspects of code content bring to light previously hidden evidence of connotations within CCEs to reinforce corporate hierarchy with a top-down, authoritarian emphasis (e.g., Winkler 2011Winkler , 2012Munter 2013). Over one-third of the content-oriented studies examine the context surrounding the codes. The papers within this subcategory of content-oriented research fall into three streams. Garcia-Sanchez and colleagues examine associations between governance structures associated with improved management behavior in the earnings management literature and CCE implementation or the level of CCE embeddedness (Garcia-Sanchez et al. 2013, 2015aRodriguez-Dominguez et al. 2009). The authors find that the greater the percentage of outside directors on the board, the greater the likelihood the corporation has adopted a code, and the greater the percentage of independent directors the broader the cope and inclusivity of the codes adopted. While the authors find significant correlations, they are unable to provide causes and effects due to limitations in methodology, namely regression analysis of archival data. As stated earlier, it is also not possible from these studies to determine the chronological order of the variables. This is an area that can be further explored through case study.
In the second stream, Svensson and colleagues utilize survey questionnaire data to examine corporations' commitment to CCEs, i.e., code embeddedness (e.g., Callaghan et al. 2012;Svensson et al. 2006Svensson et al. , 2009aSvensson et al. , b, c, 2010aSvensson et al. , b, 2011. These studies focus on differences over time or across country. Overall, the results suggest that code artifacts have increased over time for all corporations. However, adoption of code artifacts has been much slower for corporations in Sweden than for corporations operating under the Anglo governance model, and has been even slower for organizations within Sweden's public sector. The third stream of research utilizes the EIRIS database to explore whether differences in code embeddedness across countries can be explained by differences in national cultures. For example Scholtens and Dam (2007) find that differences are associated with Hofstede's four cultural indicators, while Forsyth and O'Boyle (2011) find that the greater the degree of relativism expressed by a country's residents, the less extensive the embeddedness of codes for corporations based in that country.
While most studies of code content paint a rather bleak picture wherein codes in general still appear to be inwardlooking, protectionist window dressing, findings from recent studies do however suggest that some of the larger corporations are beginning to expand the focus of the codes to be more inclusive. Weber and Wasieleski (2013) report that based on survey responses from ethics and compliance officers in the U.S. that ethics and compliance programs have become more comprehensive in scope. Stohl et al. (2009) CCE and context (embeddedness) Hofstede's (1983) dimensions of national culture Survey. Perceptual.

Not given. ≈ 2005-2007
Australia Canada USA/organization find that while the majority of CCE themes for the largest Global and U.S. corporations relate to stockholders, profits and legal constraints, the codes for these corporations also include 3rd generation ethics themes. A subsequent analysis finds the CCEs of Western European based TNCs contain a significantly greater amount of 3rd generational thinking than the CCEs of U.S. based TNCs. As such, it appears Western European corporations, at least, are beginning to expand the focus of the codes to include external shareholder considerations. Bodolica and Spraggon (2015) find that code content for most large Canadian corporations still focus on legal aspects to safeguard the firm, but about 6% of the corporations in their sample had publicly available codes that were value driven and a further 14% of corporations appeared to be working toward that end. In sum, this evidence taken together with findings that suggest smaller companies imitate larger companies' CCEs (e.g., Forster et al. 2009; Holder-Webb and Cohen 2012) provides a somewhat optimistic view for a positive evolution of CCEs going forward.

Review of Content-Oriented Papers
Country Specific, Country Comparison or Global Several studies continue the focus on country-specific features, across country considerations (by international corporations), or between country comparisons of code content. Among the studies focusing on country-specific features, O'Dwyer and Madden (2006) examine the existence, content and implementation/enforcement of CCEs by the largest companies operating in Ireland. Responses to a questionnaire reveal that issues impacting the company or the employee were widely cited, but there was a scarcity in the code content of issues impacting wider society. Specifically, the most frequently included topics in the codes relate to employee confidentiality, health and safety, protection of company information, and adherence to the law. The authors conclude that while the survey responses indicate the number of Irish companies with a formal code of ethics increased substantially between 1995 and 2000, the code content is primarily focused on issues related to company and employee protection while neglecting society protection. Weber and Wasieleski (2013) conduct a survey of corporations to also gain an understanding of the current state of ethics and compliance programs. Responses from ethics and compliance officers representing 60 U.S. companies indicate that some topics are universally covered in the codes: confidential information, conflicts of interest, employee discrimination or sexual harassment, and gifts and entertainment. The authors believe the importance given to these topics within the codes exemplifies the emphasis placed by corporations on the compliance aspects of their CCEs. The authors also conclude, however, that ethics and compliance programs, even if heavily influenced by regulatory and legal considerations, have become more comprehensive in scope. Lugli et al. (2009) perform a content analysis on the CCEs of corporations listed on the Italian stock market index, FTSE MIB. The codes were obtained directly from company websites or through correspondence with the company. The results from their analysis of frequency of mention suggest that codes place a much greater emphasis on stakeholders' behavior toward the firm than firm's behavior towards stakeholders. Further, based on the results, in aggregate, and by economics sector (industrial, financial, and service), Lugli et al. conclude "the function of the codes is to communicate the firm's position on ethical values to external parties, rather than to implement CSR internally" (2009, p. 44).
Several studies notice a convergence of corporate codes in the U.S. Holder-Webb and Cohen (2012) examine for the uniqueness of code content between U.S. firms, because Sect. 406 of SOX states that individual firms should determine the specific content of the code, and it believes codes should vary from firm to firm. The authors perform content analyses on the CCEs from a random stratified (by size) cross-sectional (across industry) sample of U.S. publicly traded firms. The researchers find that, to a large extent, the content and language of the codes is generic, and there is a convergence of code content across companies consistent with institutional isomorphic pressures. The authors suggest that the content and language of the codes are a rational symbolic response to regulation, in that, the language used minimizes the effects of the code on constraining behavior and the codes provide a decoupling of the organization's public response from the internal workings of the organization.
Similarly, Forster et al. (2009) find substantial commonalities in the textual content of U.S. firms' codes. The authors conduct textual analysis on codes downloaded from the websites of firms belonging to the S&P 500 index along with 100 firms listed in the bottom decile, in terms of market capitalization, of the NYSE and Nasdaq. The results from the analysis show that approximately 8.4% of the sentences contained in S&P 500 CCEs are duplicates of sentences contained in other companies' codes. This commonality increases to approximately 16.6% of sentences contained in the contents of the code for the small firms in the sample. While the authors expected legal-based statements to be duplicated across company code content, the findings also show there are substantial commonalities for the value-based statements. The authors conclude that smaller companies may be copying codes of larger companies to conform to regulations, i.e., there is a convergence of code content due to isomorphic pressures. Sharbatoghlie et al. (2013), in a follow-up study to Sharbatoghlie and Rezaei (2008), find a convergence of CCE content between U.S. CCEs and global CCEs across time. The authors downloaded the code of ethics from the websites of the Fortune 100 US and Fortune 100 Global companies, and performed a data-mining analysis of the content of these codes to determine the frequency of key ethical words contained in the codes. The authors find that between 2006 and 2009, the number of publicly available codes on company websites has increased from 86 to 95% for both the Fortune 100 U.S.-based companies and the Global 100 companies. In addition, a data-mining analysis revealed the differences in the frequency of key ethical word rankings between U.S.-based and Global-based companies diminished markedly between 2006 and 2009. This result suggests a convergence of code content between the U.S.-based and global-based companies.
Several studies examine whether there are differences in code content within companies across countries, i.e., international corporation codes across countries of operation. Svensson and Wood (2007) perform a content analysis on CCEs received in response to a request for codes to the top 500 Australian companies (based on revenue) listed in Business Review Weekly in 2000. The authors classify the 78 codes received based on the content of the code (uniform or flexible meaning and interpretation) and the contextual business environments for which the codes have been developed (within the same or across different environments). As such the authors classify the contents of the codes into four generic strategic approaches: standardized (corporation applies a uniform meaning and interpretation of the code within the same contextual business environment); replicated (uniform meaning and interpretation of the code across different contextual business environments); individualized (corporation assumes the need for a tailored interpretation of the code); and customized (corporation assumes the need for an adaptive interpretation of the code). The study finds that the CCEs follow a standardized or replicated strategic approach, even though most of these Australian companies operate internationally or globally.
Halff (2010) studies whether global CCEs deal with contradictions between corporate and local laws, values, customs and norms. Given the international operations of global companies, the occurrence of such contradictions is argued to be high. The sample for the study is comprised of the publicly available codes from corporations listed in the 2008 Fortune Global 500. Of the 338 CCEs analyzed, 260 (76.9%) codes do not even allude to contradictions, 25 (7.4%) refer solely to contradicting laws, and only 53 (15.7%) codes refer the broader range of contradictions between corporate law, values, customs, and practices and local norms.
Using an institutionalist perspective and drawing on a theoretical frame based on "varieties of capitalism," Amaeshi and Amao (2009) explore the embeddedness (in terms of home country influences) of codes of conduct when corporations operate outside of their national or regional institutional contexts. The study examines the content of codes of conduct of MNCs in the Nigerian oil and gas sector for exemplars of three varieties of capitalism: coordinated market economies CMEs), liberal market economies (LMEs), and hybrid, or mixed, market economies. The study finds that the codes of conduct for these MNCs tend to remain consistent with the characteristics of their home country's model of capitalism. However, MNCs from hybrid market economies tend to be more flexible in adapting their codes of conduct to the relatively weak institutional context of Nigeria than the MNCs from CMEs and LMEs.
The study by Jensen and Sandstrom (2010) perhaps sheds some light on the consistent findings of a lack of consideration to local laws, values, etc. in international corporations' codes. Arguing that corporations and researchers still need to articulate the ethical responsibilities of MNCs in a global context, Jensen and Sandstrom dissect one MNC's earnest "attempt to articulate what it means to be an ethical corporation and how it can become a global leader in ethics" (p. 282). Specifically, the authors perform a textual analysis of the contents of the Woolf Committee Report (2008) commissioned by BAE Systems Plc. The analysis reveals that the report focuses on the traditional normative (i.e., value creating) stakeholders while dismissing derivative stakeholders. In addition, the study finds that the recommendations within the report are geared towards reputation concerns while at the same time shying away from corporate responsibility. Jensen and Sandstrom conclude that rather than providing guidelines "on how a global ethics can be defined and lived" the text in the report "provides reasons for preferring and enacting a business-as-usual kind of reality" (2010, p. 290). Stohl et al. (2009) examine the degree to which TNCs' CCE content include third generation thinking about global corporate ethics and responsibilities. A content analysis was performed on the codes of 157 global corporations listed in the 2006 Global 500 or Fortune US 500. The authors find that 1st generation ethics (legal issues and shareholder profitability) comprise the largest proportion of the contents of ethics, however, three quarters of the companies did address all three generations of corporate responsibility to some degree. While no significant differences in the proportion of sections within a corporation's code of ethics addressing 2nd generation ethics (welfare of employees and their families) or 3rd generation ethics (overall social good) were found across industrial sector, the proportion of sections addressing 3rd generation responsibilities was found to be, on average, significantly higher for Western European corporations than U.S. corporations. The authors conclude that while the focus of the codes still reflect a concern for corporate profitability and compliance with the law, the codes are beginning to reflect an expanded view that considers external stakeholders. Grein and Gould (2007) offer a unique perspective on how codes may be implemented with greater success in the global arena. The authors propose that, for multinational corporations, a code of ethics which addresses group membership salience and is coordinated across markets will be viewed more positively and generate greater compliance than one that does not. Taking a marketing approach to developing and integrating codes into global markets, the authors argue that global integrated marketing communication concepts can be used to promote the ethical values contained in the codes with group membership salience providing a focus on the content. Green and Gould compare four codes against their propositions and conclude "relatively little attention is being paid to marketing issues, and there is little which links directly with globally integrated marketing programs" (p. 300) which can result in legitimacy gaps and code failures. In essence, the basic premise of the authors' proposition is that a company's code of ethics values should be promoted in a manner similar to corporate branding which will lead people to associate the company with the promoted values, and in identifying with the company the individual will need to hold, or uphold, those values.
A few studies compare codes for similarities and differences in content between countries. Winkler and Remišov (2007) examine the code of ethics from 41 large German firms and 23 large Slovakian firms for content similarities and differences based on a set of 20 common ethical issues (e.g., respect for the law, conflict of interest). The study finds that the codes of Slovak firms tend to be internally-oriented with a focus on issues like commitment from employees and protection of company resources. In contrast, the codes of German firms tend to be externally-oriented, focusing on issues like environmental protection and anti-discrimination. The authors reason that the main driving force behind the differences in the ethical issues raised in CCEs between the firms of each country is the legal, political, and economic transformational processes at play in Central and Eastern Europe compared to the relative stability in Western Europe. Based on their results, the authors conclude that firms do consider both internal and external situational factors when developing CCEs. Melé et al. (2006) in comparison find a similarity between countries when examining code content for companies based in Argentina, Brazil, and Spain. The similarity is interesting because it is counter to findings from studies of companies based in other countries. A questionnaire consisting of 16 items was sent to the largest commercial companies in each country during the early 2000s. Responses indicate that for companies in all three countries the primary motive for having a code is to express and spread corporate values and corporate philosophy. Specifically, respondents indicated that the most important principles applied in drafting the code are human virtues, universal ethical principles and generally accepted values-utilitarianism was a distant fourth. Melé et al. (2006) note that respect for the law is not as strong in Latin countries as it is in Anglo-Saxon and Scandinavian countries, and that people in Latin countries "are more concerned with legitimacy than with legality" (p. 28). Data for Mele et al.'s study was collected between 2000 and 2002, as such, it would be interesting to examine whether the findings still hold today.

Changes Across Time
Several studies focus on changes in the content of codes across time. Canary and Jennings (2008) utilize centering resonance analysis to identify preand post-SOX similarities and differences in the content of CCEs. The sample for the study consists of CCEs for 23 companies, of which, 22 have international operations and 22 are based in the U.S. The study finds that the set of codes for 18 of the 23 companies in the sample exhibit a low resonance, indicating a significant change in the content and structure of the codes pre-and post-SOX. Additional analysis reveals a greater emphasis on law and compliance post-SOX, but also the emergence of the word "ethics" as a value or principle in post-SOX code content. Singh (2006), in a follow-up study to Lefebvre and Singh (1992), examines similarities and changes in the code content for Canadian companies between 1992 and 2003, by conducting an analysis of the frequency of mention for 61 items across seven categories. The comparison of code content between 1992 and 2003 revealed a significant increase in the mention of "legal responsibility" as the basis of the codes, as well as an increase in the frequency of mention of code enforcement and compliance procedures, environmental affairs in conduct on behalf of the firm, and environmental laws. Consistent with the results from the 1992 study, the codes are mostly concerned with actions against the firm than actions by the firm. Singh et al. (2011) replicate Singh et al.'s (2005) study to examine longitudinal changes in the content of Australian, Canadian, and Swedish corporation's code of ethics. Consistent with Hofstede's dimensions, cross-culturally, the CCEs in Australia and Canada are very similar for all categories and are more prescriptive and regulatory than those in Sweden. While these results are consistent with Singh et al. (2005), the codes of the Australian and Canadian firms were also found to have become more prescriptive over time which intensified the differences in code content compared to Swedish firms.

Industry Specific
There are only three industry-specific CCE content analysis studies. Ki and Kim (2010) perform a content analysis of the 605 CCEs and ethics statements that were publicly available on the websites of 1891 U. S. public relations firms in 2007. The authors find that public relations firms tend to focus on advertising the firm's services rather than presenting ethical values, and the values least mentioned are those highlighted in the Public Relations Society of America's Code of Ethics. Oehmen et al. (2010) examine supplier codes of conduct in the electronics industry (discussed further in the Supplier Codes subsection below) and Munter (2013) examines the symbolic aspects of CCEs in the Swedish banking industry (discussed further in the Language sub-section below). Although the focus of the research differed between each of these studies, one common theme within the findings is that the code content is based on providing self-serving purposes.

Supplier Codes
Several studies examine MNC codes of conduct for suppliers. Preuss (2009) conducts a content analysis on 44 CCEs, obtained from FTSE 100 corporations as listed in 2007, that specifically address CSR issues in the company's supply chain. The content analysis is based on frequency of mention of items within three categories: environmental, employment, and economic. Pruess finds that labor standards for supplier employees are the most frequently mentioned issues in the code, followed by some general environmental issues at supplier plants, while the economic issues for the supplier are mostly ignored. The employment issues in the codes reflect the contents of the Ethical Trading Initiative (ETI) Base Code which, not coincidentally, ETI members are expected to adopt. As such, the employment criteria contained in the codes are homogenous. The results from the study lead Preuss to conclude that the motivation for large companies to adopt ethical sourcing codes appears to be a response to public criticism, anticipated legislation, or perhaps a means to gain a competitive advantage, but they find little support for adoption due to organizational ethical values and principles.
Following a similar technique, Preuss (2010) examines the content of codes at the organizational and sub-organizational level for FTSE 100 corporations. The sample consists of 281 codes: 77 company-wide codes and 204 sub-company codes that address six particular corporate functions (e.g., environmental, procurement, ethical sourcing). He finds that the hierarchy of codes, when taken in conjunction with supra-organization (e.g., national, industry, NGO) level codes, form a "lattice-work of codes [which] allows firms to convey different messages to different audiences" (2010, p. 484). Specifically, the visible organizational level codes contain generalized stipulations and vague commitments, whereas the sub-organizational codes contain specific stipulations and requirements of firms within the supplychain. Based on the findings from this study, He concludes that large corporations not only present codes as windowdressing for the general public, but these large corporations also coercively pass responsibility for ethical, environmental and social issues down the supply chain through internal sub-organization level codes. Oehmen et al. (2010) conduct a content analysis of supplier codes of conduct (SCoCs) adopted by 24 (of 75) companies in the electronics industry as listed in the 2007 Forbes Global 2000. The study determines the percent of codes that include each of 68 supplier code issues grouped into 5 categories: labor standards; health and safety issues; environment issues; ethics issues; and compliance. The frequencies of inclusion for each of the items within these categories tend to suggest a legal protectionism viewpoint. Issues related to labor standards receive the greatest coverage while issues relating to the company's impact on society receive little attention. While some companies' codes include general compliance issues, these codes rarely provide specific details on processes and procedures. Schwartz (2005), building on the study by Gaumnitz and Lere (2002) attempts to identify a set of core universal moral values for a corporate code of ethics, i.e., moral values that "retain their significance despite differences in culture, religion, time, and circumstance" (Schwartz 2005, p. 31). To define the set of universal moral values, Schwartz performs a comparative analysis of moral values from three sources: companies' codes of ethics, global codes of ethics, and the business ethics literature. The author concludes that there are six universal moral values: trustworthiness, respect, responsibility, fairness, caring, and citizenship. Interestingly, these same six values are identical to the six core ethical values of the 1992 Josephson Institute of Ethics Aspen Declaration. Donker et al. (2008) examine whether there is a positive association between code of ethics quality and the economic performance of publicly traded companies in Canada. The measure of code of ethics quality is based on the number of commonly accepted positive ethical values that are included in company's code of ethics. The ten ethical values that make up the self-generated index set build from the values listed in the Schwartz study. The results from a regression analysis reveal that a firm's code of ethics quality is positively associated with its market to book (MTB) ratio after controlling for firm size, financial leverage, return on market capitalization, and block shareholders. Further analysis reveals that the initial results are driven by firms with MTB ratios of less than one. In other words, the results suggest code of ethics quality does not impact market perceptions of future firm performance for high growth firms. Bodolica and Spraggon (2015) analyze publicly available codes of corporations listed on the Toronto Stock Exchange (TSX) that had initiated a merger and acquisition within the previous decade. Merger and acquisition companies were chosen because of their relatively larger size, greater diversification, higher visibility, and greater influence over 1 3 corporate stakeholders. Of the 237 corporations identified that fit the criteria, 131 had publicly available CCEs. The content analysis consisted of a word search based on the six general aspects recommended by TSX corporate governance guidelines and the six moral universal standards suggested by Schwartz (2005). The researchers suggest that corporations can be classified into one of four clusters of behavior based on the results from the analysis. The 106 firms which did not have, or did not have publicly available, codes are classified as "shadowers." Of the 131 firms with publicly available codes: 84 (64.1%) firms are classified as "compliers" because code content and adoption "is driven mainly by the corporate desire to adhere to legal standards for better safeguarding the firm;" 33 (25.2%) corporations are classified as "updaters" because their codes are continually being updated as the firm strives to move from a "complier" to a "promoter" however, the codes "still lack a well-developed element of social values;" and only 14 (10.7%) firms are classified as "promoters" because the codes are "essentially value-driven, where the most sophisticated values of caring and social responsiveness are emphasized" (Bodolica and Spraggon 2015, p. 471).

Ethics Quality
Utilizing a different operational definition of code of ethics quality, Garegnani et al. (2015) examine whether the quality of codes adopted by Italian corporations is related to firm size, degree of internationalization, and the industrial sector in which the firm operates. The measure of CCE quality is a self-generated index based on 40 indicators from the following six equally weighted categories: "commitment from the top, style and availability, whistle-blowing, relations with stakeholders, compliance procedures and legal items" (p. 547). Results from Tobit regression analyses reveal that code of ethics quality is positively and significantly related to firm size (market capitalization), but surprisingly, code of ethics quality is negatively and significantly related to the degree of internationalization (ratio of foreign to total revenues). The control variables of profitability, financial leverage, and primary shareholder influence also have a significant negative relation to code of ethics quality. Garegnani et al. conclude that management of larger firms "are better able to focus on the formal aspects of their codes of ethics and on their readability and dissemination" (p. 552), however, management for firms of all sizes may only view codes as applicable within the confines of home country culture.

Language (Symbolic Aspects)
Several studies examine the use of language in the code rather than just the words contained in the code. Béthoux et al. (2007) download from the 2004 ILO database, and perform lexical analysis from a CSR perspective on, 175 CCEs and framework agreements adopted by 166 MNCs. The analysis provided three main themes in this corpus of codes: respect for ILO norms; com-pany's relationship to society; and, internal discipline and organization. Findings from the study reveal that the main focus of the codes is on employee compliance and protection of assets, with little focus on corporate responsibility.
An analysis by Long and Driscoll (2008) of the semantics of CCE content for the largest corporations in Atlantic Canada reveals that the codes are implemented for strategic legitimacy. This strategic approach to legitimacy is a retrospective process involving cost-benefit analysis whereby managers seek to give the appearance of close alignment with society's current values. As such, the authors conclude the resulting codes are insincere and distanced from moral foundations. However, while the authors present a detailed study of the overall tone of the code, the analysis is based on a very small (n = 7) and geographically localized sample.
Winkler (2011) utilizes critical discourse analysis on the CCEs adopted by German Dax30 companies to investigate how internal actors are positioned in terms of attributes ascribed to them and the defined relations between them. Specifically, the study examines how the language used by the code's authors influences internal actor perceptions of corporate reality. Four categories of internal actors were identified in the codes: employees, line-managers, top management, and compliance officer/board. The analysis reveals that the roles of each group follow the existing management ideology and company hierarchy. Specifically, employees are "passive receivers of code instruction" (p. 658), line-managers are the code implementers and enforcers, top management represent the company and are the initiators, authors, and authorizers of the code, while the compliance officer or compliance board, is placed just below top management in the hierarchical chain (which also serves to increase the distance between top management and line-managers and employees) and is given the legitimacy to implement, interpret, and enforce the code.
Winkler (2012) builds on Winkler (2011) with an examination of the discursive practices employed in CCEs adopted by German Dax30 companies to determine how the language of the code categorizes and manages the self-identification of employees and their position within the company. The study finds that the discursive practices contained in the codes position employees into four different identities: the equal employee (part of the community), the responsible employee (compliance with the code), the self-monitoring employee (commitment to the company), and the subordinated employee. The first three identities are prevalent in sections of the code that emphasis egalitarianism, deemphasize hierarchy and status, and regard all groups as homogenous, e.g., the language in the introduction section of the code provides a sense of community to legitimize the code and to emphasis a shared sense of responsibility. The subordinated employee identity is prevalent in sections related to behavioral rules and regulations as well as code implementation, compliance and enforcement. The study shows that the symbolic aspects of language within CCEs are used to position and control employee identity within the organization. Munter (2013) argues that "codes of ethics constitute corporate acts towards the employees that are not, per se, beyond questions of ethics" (p. 174), i.e., corporations have ethical obligations to employees. As such, he performs a content analysis with a normative ethical perspective to determine how the codes address and treat employees. The analysis draws on the values of respect, equality, reciprocity and care from the HRM ethical management practices literature to determine if code language treats employees unfairly (e.g., demeans, disrespects, treats as just another resource, or emphasizes unequal positions in the corporate hierarchy). The sample for the analysis consists of nine codes obtained in 2006 from the 18 banks that are members of the Swedish Bankers' Association. Munter finds that eight of the nine codes examined are in conflict with these values, with the common ethically problematic design characteristics being: lack of ethical value-based reasons given; authoritarian tone and imbalanced focus; and unreasonable freedom restrictions. Consistent with prior research the findings suggest that the codes take a top-down perspective, emphasis unequal hierarchical positions, and are used as a tool for control, conformity, and compliance. Hoover and Pepper (2015) analyze the content of the codes of the companies listed in Fortune magazine's 2009 "100 Best Companies to Work For" to identify the extent of ethical frameworks utilized in the statements. Specifically, the authors examine the 93 publicly available codes for evidence of the use of three ethical frameworks: deontological, teleological, and ethics of care. The deontological framework was found to be the most prevalent, with 77 (83%) of the codes containing deontological references. In comparison, the ethics of care framework was identified in 63 (69%) of the codes, and 51 (55%) of the codes contained teleological references. The majority of codes (70%) contained multiple frameworks, with the remaining 28 (30%) single framework codes being predominantly deontological. A subsequent analysis using dictionary of affect in language (DAL) software shows that codes containing ethics of care references were significantly different in terms of the pleasantness of the statements compared to codes that did not. Interestingly, organizations with codes containing ethics of care statements were also found to employ a significantly higher percentage of women.

CCE Support Structures
Several studies examine business ethics support structures surrounding the codes, e.g., governance structures, code embeddedness, and culture. Garcia-Sanchez and colleagues utilize archival data to explore the effects of Board of Director and CEO characteristics on the likelihood of CCE adoption and the extensiveness of codes if adopted. Rodriguez-Dominguez et al. (2009) test whether a greater percentage of outside directors on the Board, a greater percentage of stock ownership by the board members, and a greater percentage of women directors on the Board, is positively related to the likelihood that the company has a CCE on a sample of 351 corporations from Spain, Italy, and the UK. The percentage of outside directors and percentage of stock ownership were found to be significantly positively related to the existence of CCEs in Spain and Italy, but all other hypothesized relationships were rejected. Garcia-Sanchez et al. (2013)  In addition, the authors test whether the type of governance structure (Anglo, Germanic, or Latin) moderates the influence of board involvement in ethical issues. The authors find that the number of independent directors is positively related to the likelihood of CCE implementation in all environments, but the number of women directors only increases the likelihood for companies under Germanic governance structures. The results from this study also indicate that the higher the number of independent directors and the higher the number of female directors, the broader the scope of the CCEs implemented by the company. However, the level of involvement by independent board members is conditioned by the corporate governance system, with independent directors in the Anglo context exhibiting more influence.
In a separate series of papers, Svensson and colleagues utilize survey data to study corporate commitment to CCEs by examining the extensiveness of ethics program support structures, i.e., the embeddedness of the CCEs. This series of studies tends to focus on comparisons of corporate ethics artifacts across time and across countries. Svensson et al. (2006) find that in 2002, only 56% of top 100 corporations by revenue in Sweden had a code of ethics, and the authors suggest that those having a code exhibit a lack of commitment to supporting business ethics due to the self-reported under-utilization of ethics committees, ethics training committees, ethics training, ethics ombudsman, ethics audit, whistleblower protection procedures. Similarly, Svensson et al. (2009a) find that only 27% of Sweden's largest public sector organizations across the three categories of government, county councils and municipalities had a code of ethics during their survey periods 2001-2002-2006. Svensson et al. (2010a in a replication study of Svensson et al. (2004) compare business ethics commitment between private and public sector organizations in Sweden, and find that over the time between the two studies, CCEs within the public (government) sector developed to a lesser extent than those in the private sector. They also conclude, however, that support mechanisms associated with business ethics commitment are still being under-utilized in all sectors. Singh (2011b) replicates Singh's (2006)  Building on their prior research, Svensson and colleagues propose models of code of ethics embeddedness. Svensson et al. (2009b) utilize survey responses from Swedish corporations and public entities to perform an exploratory factor analysis of fifteen items across five dimensions to provide a measure of the embeddedness of a code of ethics. Svensson et al. (2010b) perform confirmatory factor analysis and structural equation modeling to test a four construct (15 indicator variable) measurement model utilizing the data from Svensson et al. (2009b). Similarly, Svensson et al. (2011) using survey response data from the top 500 corporations in Australia, Canada, and the U.S., perform confirmatory factor analysis and structural equation modeling to validate a fourfactor model of the embeddedness of CCEs building from the exploratory factor analysis of the same data in Svensson et al. (2009c). Data from corporations in Australia, Canada and the U.S. were chosen because of the cultural closeness of these three countries. The confirmatory factor analyses reveal four factors (surveillance/training, internal communication, external communication, guidance) are represented by eleven indicator variables. The authors report that the model meets the recommended requirements for goodness of fit, construct reliability, as well as convergent, discriminant, and nomological validity.
Taking a slightly different tack, Callaghan et al. (2012) also use survey responses from the top 500 corporations in Australia, Canada, and US, to examine a model of code of ethics embeddedness involving nine measures across two dimensions: staff support (ethics audits, ethical performance appraisal, and consequences for a breech) and regulation (ethics committee, ethics training committee, ethics training, ethics ombudsman, whistleblower protection, and guide to strategic planning). The analysis revealed that companies in the U.S. ranked highest in frequency for each of the artifacts except for the use of the code to guide strategy. The authors note that the U.S. Sentencing Guidelines cover all of the items except for the strategic planning aspect, and conclude that companies in the U.S. have the correct window-dressing but "shaky" foundations. Utilizing this same model, Mpinganjira et al. (2016)  In a separate study, Oladinrin and Ho (2016) examine possible factors that complement successful CCE implementation in the construction industry in Hong Kong. From a convenience sample, the authors utilized a questionnaire survey to gather construction industry practitioners, senior managers, and supervisors' opinions about code of ethics context attributes within their organizations. Specifically, the questionnaire incorporated 30 items designed to capture attributes of code implementation based on the five "enabler" criteria of the European Foundation for Quality Management (EFQM) excellence model (leadership, processes, people/employee, policy and strategy, and partnership and resources). Exploratory factor analysis and structural equation modeling analysis of the response data revealed that 26 of the items formed five factors closely related to the EFQM model. Based on the results from their analyses, the authors suggest that while leadership support and commitment is important to CCE implementation, ethical concerns must be addressed in an organization's strategy and supported by policies and procedures for effective CCE implementation in the construction industry. Although the context between this study and the study by Callaghan et al. (2012) differ, the results from this study do provide some support for Callaghan et al.'s conclusion about the importance of CCEs in guiding strategic planning.
In an across country study, Aydinlik et al. (2008) compare the embeddedness of ethics codes between the top 500 Turkish and the top 500 Swedish corporations. Aydinlik et al. (2008) find that Turkish companies may be more advanced in installing attributes to "advance the ethos of codes into their organization" (p. 785) than Swedish companies. Specifically, Turkish companies were found to be more likely to have the following ethical artifacts: ethics committee, ethics training committee, ethics training, ethics ombudsman, employee ethical performance appraisal, and whistleblower protection procedures, The authors believe the across countries differences in the implementation of ethics artifacts in corporations can be partially explained by cultural differences, in that, Turkish companies follow the prescriptive Anglo-Saxon cultural model, whereas Swedish companies follow the laissez-faire Nordic cultural model. It should also be noted, however, that only a small number of Turkish companies had implemented codes (32 of 137 respondents).
A few studies try to explain differences in code of ethics embeddedness across countries based on national cultures utilizing data from the Ethical Investment Research Service (EIRIS) database for their analysis. Scholtens and Dam (2007) examine the data from EIRIS to determine whether ethical policies of firms differ by country or industry. The sample from the database consists of 2681 companies from 24 countries and 35 industries. The authors generate a measure of code of ethics extensiveness by assigning a numerical scale to five EIRIS measures related to the code of ethics: governance of bribery and corruption; systems (existence and comprehensiveness) of the codes; implementation (comprehensiveness of implementation system) of the codes; communication of the code; and human rights policy. Consistent with prior research, but on a broader scale, the results show significant differences in corporate code of ethics policies between countries for all five measures of code of ethics extensiveness. However, the authors are unable to discern whether the differences are driven by industry or country.
Using data from Hofstede's (1980Hofstede's ( , 1991 studies, Scholtens and Dam (2007) also perform a rudimentary test of the association between differences in ethics policies across countries and Hofstede's four cultural indicators: power distance, individualism, masculinity, and uncertainty avoidance. The results suggest that individualism is significantly and positively associated with CCE comprehensiveness and human rights policy, whereas uncertainty avoidance is positively associated with all the CCE variables except for human rights policy. In contrast, power distance and masculinity have a predicted negative association with all five code of ethics variables, although only the human rights policy is significant.
Forsyth and O'Boyle (2011) examine whether the extent to which organizations rely on CCEs is related to the moral beliefs of its country's residents. Ethics Position Theory (EPT) states that a person's ethics position, which guides moral judgment and behavior, differs across two dimensions: relativism (skepticism of guiding principles to define right and wrong), and idealism (concern for outcomes that cause harm). The authors focus their study on 11 countries whose businesses were included in the EIRIS database as reported in Scholtens and Dam (2007) and whose residents had been previously studied for ethical position (degree of relativism and idealism). The five EIRIS indices related to CCEs provide a measure of the code extensiveness for businesses within each country. As predicted, the results show that the greater the degree of relativism expressed by a country's residents, the less likely are businesses in that country to develop extensive CCEs. The authors, however, find no support for a relationship between degrees of idealism and code extensiveness.

Output-Oriented Studies
Approximately, 35% of the papers reviewed are classified as output studies and further organized under three subcategories due to the research question in focus. Output studies continue to focus on the behavioral effects of CCEs with about 45% of the output studies reviewed focusing on effects at the individual level. A much smaller group of studies focus on the effects of CCEs on output at the organizational level. A new subcategory, which accounts for about 35% of the output studies, examine effects in the supply chain, where the empirical context often encompasses labor conditions in South-Asian factories (see also transformation studies).
The golden question of whether or not CCEs are effective arguably has the most plausible chance of being answered by studies within this category. The overall answer to the question is yes, with a majority of studies in this category showing somewhat positive effects of CCEs within the studied context. This is partly in contrast to the review reported by Helin and Sandström (2007), but consistent with Stevens (2008). Based on her review, Stevens (2008) argues that the question now is not if CCEs have an effect but rather what makes them effective. Her stance fits well into the somewhat mixed picture gleaned from the output-oriented studies reviewed in this paper.
However, the context examined as well as the operationalization of variables differs widely across studies, which makes generalizations difficult. Specifically, while the CCE is always one of the antecedents, the operational measure of CCE varies across studies. In addition, there is no consistency across studies for antecedent variables examined in conjunction with CCEs, nor for variables measuring consequences. Even within the same study, different contexts and multitudes of variables are measured which results in some positive effects, some negative effects and some effects that are not statistically significant. Table 2 provides a list of the papers reviewed under the output-oriented category including author(s) and publication date, subcategory, theory, method, data collection date, context and level of analysis, antecedents, and consequences. The table is followed by an overall summary of the papers reviewed within this category, and a review of the individual papers by subcategory.

Summary of Output-Oriented Studies
This category reviews 36 studies: 12 of the studies focus on ethical output in terms of labor conditions in the supply chain, 16 on ethical output in terms of behavior on an individual level and eight studies on other, mostly organizational level, output consequences. Studies on ethical output provide an optimistic picture of the effects of CCEs. However, the image is fragmented, as the concepts and variables used differ considerably between studies.
The empirical focus as well as level of analysis differ between the categories. Studies within the effects on supply chain/labor condition subcategory, typically focus on CCE effects in conjunction with local management at the location and/or the local rules and regulations, i.e., the codes at the supplier location of the buying company (e.g. Bartley and Egels-Zandén 2015). In the second subcategory, effects on ethical output (individual level), the studies explore how ethical output is affected in companies at a personal/individual level (e.g., Petersen and Krings 2009;von der Embse et al. 2010). Studies within the third subcategory, effects on other output variables (organizational level), report on other output effects of CCEs at an organizational level (e.g., Erwin 2011). This last subcategory is fragmented into several focus areas, such as, the role of philanthropy as a mediating variable between codes and job engagement (Lee et al. 2014), or whether codes are a predictor of CSR performance (Mijatovic and Stokic 2010). While studies scrutinizing labor conditions in supply chain locations are mostly performed in emerging market countries like China, Malaysia, South Africa and Turkey.(e.g., Egels-Zandén 2007; Yu 2009), most of the studies in latter two subcategories are administered in the Western world in countries like the U.S., Canada, Germany and Spain (e.g., Arnold et al. 2007).
Case studies are rare but some of the labor condition studies are carried out as case studies with the collection of data through interviews or observations along with the collection of documents. Only a few of these studies are performed inside factories (e.g. Yu 2008Yu , 2009Yu , 2015, but the majority of case studies collect data indirectly through interviews with union representatives or auditors gathering perceptual data which possibly could lead to biases in the data. Within the supply chain/labor condition subcategory, a few studies utilize surveys of purchasing departments in the buying companies. In contrast, survey questionnaires with focus on perceptual data is the dominant data collection method for studies in the subcategory ethical output at the individual level. Studies of output at the organizational level also utilize surveys to collect data, but there are also some document studies of actual data in this subcategory. The data collection date is seldom mentioned in the output-oriented studies but most of the data is collected before 2010. From this limited data, there are no obvious time-related trends. The CCE is the antecedent in all studies. For studies in the supply chain/labor conditions subcategory, the main antecedent is the code of the buying company and the consequences are what happens at the supplier-end, usually at a factory. CCE as the main antecedent is combined with other variables such as local laws, government labor inspections, external monitoring, compliance pressure, and economic pressure. For studies of output at the individual level, CCE is combined with other variables or contextual aspects such as communication, training, trust, transparency, respect, loyalty, embeddedness, conflicts, sanctions, integration, management experience, ethics performance appraisal, various elements of ethics programs, individual beliefs, enforcement of CCE, and idealism versus relativism (e.g., Petersen and Krings 2009;Valentine and Johnson 2005). Results also indicate that scope and composition of programs in relation to CCEs are important antecedents for mitigating unethical behavior (Kaptein 2015), and that CCEs without management interaction are less effective (Kaptein 2011). Studies examining effects at the organizational level, the more common antecedents operationalized in conjunction with CCEs are, control mechanisms, code content quality, philanthropy, accreditation, and board characteristics (e.g., Erwin 2011; Lee et al. 2014). Despite the diverse variety of antecedents examined across studies, the overall direct and indirect effects of CCEs on output at the organizational level can be interpreted as positive, e.g. increased sales volume (Gallego et al. 2016), reduced employee turnover intention (Lee et al. 2014), and higher CSR ranking (Erwin 2011).
The results (consequences) from studies within the supply chain/labor conditions subcategory provide support for CCEs having a positive effect on labor conditions but there are also a couple of counteractive results. The main finding in this subcategory of studies is that the effects of CCEs are conditioned. For example, positive effects may occur but require top management commitment (Withers and Ebrahimpour 2013), or are dependent on consumer pressure in the purchasing country (Koçer and Fransen 2009). Studies reporting counteractive effects indicate that promoting CCEs in a context of global agreements may have a counteractive rather than an additive effect (Egels-Zandén and Hyllman 2007), and promoting CCEs could start a "race to ethical and legal minimum" (Yu 2008).
Studies examining the effects on ethical output at the individual level define and measure ethical behavior in a wide variety of ways which makes an overall analysis of the results less robust. For these individual-level studies, ethical perceptions, intent, evaluation, attitudes and decision-making are used and can be understood as more or less representing the same concept. Ethical perception is used in a handful of studies (Arnold et al. 2007;Singh 2011a;Valentine and Johnson 2005;Vitell and Hidalgo 2006;Withers and Ebrahimpour 2013) but across these studies ethical perception is related to different antecedents and theoretical underpinnings. In other studies ethical output is described as (un)ethical behavior (Kaptein 2011(Kaptein , 2015, internal significance (Frostenson et al. 2012), common (ethical) boundaries (Tjosvold et al. 2009), incorruptibility (McKinney and Moore 2008) and perceived ethical practice (von der Embse et al. 2010). Consistent with studies examining effects on ethical output at the individual level, studies examining effects on output at the organization level also only allow for a fragmented picture to be derived due to the wide spectrum of studies. Within this subcategory, the articles examine the effects of CCEs on sales volume, job engagement, customer trust and CSR (three studies). However, in almost all of these studies there is an indication of a positive effect on the dependent variable.
The use of explicit theory is rare in the output-oriented studies reviewed. Among the studies building on some theoretical base, new institutional theory (Egels-Zandén 2007, 2014 and agency theory (Hoang and Jones 2012) are the prevalent theories utilized in the supply chain/labor condition studies. Among the studies examining ethical output at the individual level, theory varies from cultural constructs (Arnold et al. 2007) to decoupling (Frostenson et al. 2012). In the last subcategory, effects on output at the organizational level, no explicit theoretical underpinnings are present.
To sum up, a majority of the output studies point towards a positive effect of CCEs. The results in terms of labor conditions, show mixed results; there are indications of positive effects but also counteractive results. The main finding in this category of studies is that the effects of CCEs are conditioned. Positive effects may occur but requires for example top management commitment (Withers and Ebrahimpour 2013), or are dependent on the pressure from the consumer side in the purchasing country (Koçer and Fransen 2009). Furthermore, studies on ethical behavior, or similar, show a more optimistic picture of the effects of CCE than previous studies. However, the image is fragmented, as the concepts and variables used differ between studies.

Review of Output-Oriented Papers
Effects on Supply Chain/Labor Conditions The first subcategory consists of studies focusing on labor conditions in the supply chains of mostly Western companies. Many of these studies are carried out close to supply chain production facilities where Western company codes are intended to have a positive effect on environmental and social concerns. Generally, even though the studies focus on diverse aspects of labor conditions and workers' rights, codes seem to have positive effects but these studies also show that codes can sometimes have counteractive effects.
Egels-Zandén and colleagues carry out a handful of studies on labor conditions related to CCEs in supply chain factories located in East Asia. Egels-Zandén (2007, 2014 apply new institutional theory to a longitudinal case study of Chinese suppliers in the toy industry. Egels-Zandén (2007) conducts unannounced interviews with 108 employees outside of Chinese supplier factories in 2004. The findings reveal a lack of compliance with the MNC's CCEs in the suppliers' operations. Based on the results from the study, the author provides two main explanations for this lack of compliance. First, the suppliers, through deceptive practices, managed to decouple the monitored and operational parts of their organizations. Second, the MNC, supplier, and employees have varying economic incentives that diminish the incentives to comply with the code. Egels-Zandén (2014) revisits the four factories still existing some 5 years later with the aim of finding out if CCEs over time improve workers' rights. The results, from interviews with over 100 employees outside the supplier factories in 2009, indicate that the suppliers had recoupled the monitored and operational parts of their organization. As such, and in contrast to previous studies, the results suggest codes substantially improve workers' rights over time. Two key drivers of this "recoupling of policy and practice" (Egels-Zandén 2014, p. 68) are increased monitoring and increased external pressures on the supplier. Egels-Zandén and Lindholm (2015) follow the same track in a study of factory audits in 43 garment factories located mostly in developing countries, to examine whether CCEs improve working conditions for the workers. Findings from the study provide some support for codes improving workers' conditions overall, but show no statistically significant improvements for any one specific area of the code. The authors find it surprising that, due to the limited improvement effects, companies invest huge amounts of money in codes and audits. They conclude "that this contradiction is explained by either an illusion of improvement […] and/ or pre-first-audit improvements and/or that codes mitigate overall compliance decline" (Egels-Zandén and Lindholm 2015, p. 38). Bartley and Egels-Zandén (2015) analyze data from face-to-face interviews with union officials at 192 factories (apparel/textiles, footwear and electronic) located in Indonesia. Links between codes of conduct and labor conditions detailed in government labor inspection documents were also studied. The researchers examine whether CCEs and government regulation impact working conditions across three areas: labor union rights, health and safety, and employment practices. Results from the study suggest CCEs improve areas of health and safety standards compliance, while government regulation has a positive impact on some employment practices. Neither type of regulation, private or public, was found to have an impact on labor union rights. The authors attribute the differences in level of impact to four factors: interest of the consumer public; challenge to managerial control; relative cost to implement; consistency with a technocratic compliance perspective. As such, Bartley and Egels-Zandén (2015) conclude that codes are not always meaningless but "their significance depends on the issue at hand" (p. 37), and that CCE and government regulation have complementary roles.
Egels-Zandén and Hyllman (2007) conducted a case study on a Swedish transnational company with operations in a developing country to compare two different approaches addressing responsibility for workers' rights; CCEs and global agreements. The researchers study how CCEs and global agreements interactively, as well as independently, affect workers' rights. The results indicate that workers' rights addressed in CCEs are also included in global agreements. Furthermore, the study indicates that to promote CCEs in the supply chain has "negative interactive effects on global agreements" arguing that "the current focus of codes of conducts is counterproductive for the promotion of workers' rights" (p. 207).
In the same vein, Koçer and Fransen (2009) conduct case studies with interviews in three Turkish textile companies in order to find out if CCEs help or hinder the workers position in emerging economies, with a particular focus on freedom of association. The findings indicate that the effect of CCEs is conditioned by many factors such as relationships between buyer and supplier, supplier strategy, domestic or foreign market orientation, content of the code, and consumer sensitivity. The authors argue that codes could have a positive impact on freedom of association under the very specific condition that "a partly foreign-owned and entirely exportoriented supplier establishes subcontract links with a small number of main firms which offer long term partnerships and adopt multi-stakeholder codes of conduct, while aiming to sell their final products in places where consumer sensitivity for workers' conditions is high, and where a nonopportunistic trade union organizes the supplier's factory" (p. 252). Yu (2008Yu ( , 2009Yu ( , 2015 reports on a case study conducted during 2002-2005 at one of Reebok's major footwear supplier factories in China. Examining the social impact of CCE implementation, Yu (2008) reports a counteractive effect, specifically "a 'race to ethical and legal minimum legal standards when notoriously inhumane and seriously illegal labor rights abuses were curbed" (p. 513). The workers were forced to work faster and harder, earned less, and the union was more of a company union than workers' union. Structural forces and agency-related factors were used to explain this counteractive effect of CCEs. Explanations for the ineffectiveness of Reeboks codes include tensions between the company's profit maximization goals and commitment to workers' rights, the competitive reality of the marketplace, and insufficient national regulation protecting labor rights. Yu (2009) explores the dynamics and performance of worker participation in the implementation process of Reebok's code of conduct. The study finds that although worker participation seems to have improved code implementation, the management style remains authoritarian and workers are caught in a three-player power structure. Specifically, the paper finds that various types of worker empowerment programs implemented by Reebok at one of its major suppliers have enhanced worker participation in the code implementation processes. Complaint boxes allow for individual worker monitoring and play a communicative role. Worker-elected trade unions allow for collective participation and play a communicative role and to varying effectiveness a consultative role. However, the union's effectiveness is found to be dependent on a three-player power structure-Reebok, ACFTU (The All China Federation of Trade Unions), and the supplier firm's management.
Similarly, Yu (2015) examines the effectiveness of CCEs in improving labor standards in a global production setting. The result of the study "sheds light on the influences of a variety of related factors that are embedded in overlapping international, industrial, national and local contexts to contribute to a more comprehensive understanding of the effectiveness of codes of conduct at curbing labor abuses." (p. 167). The study provides a broadened perspective on CCE effectiveness rather than giving a simple yes or no-answer to whether the code is effective. Hoang and Jones (2012) conduct interviews with workers, union representatives and managers at three factories in Vietnam in 2008 and 2010. The authors intend to clarify the ineffectiveness of CCEs, underpinned by a network perspective. The intentions of CCEs in a MNC context are argued to be control of subcontractors in a typical principal-agent model. However, the authors argue that supply chains function in a network-oriented cooperation mode. Furthermore, the MNC, factory workers, and intermediary workers, all have a common interest in violating the code. CCEs, it is argued, may, however, be helpful in aspects like improving labor regulation.
With a focus on the purchasing side of business, Withers and Ebrahimpour (2013) conducted a global survey with a sample of 5000 small, medium and large enterprises in 12 industries (response rate 8.5%) in order to study the effects of CCEs on the supply chain. The findings indicate that the range of influence is fairly limited to managers and employees within their own company; "sharing CCEs with members of ones' supply chain has virtually no positive impacts on the behaviors of vendor or B2B customer employees" (p. 37). Within the own company, the study reports that formality of CCE, as well as top management and employee commitment, has a positive impact on behaviors and perception.
In the same vein, drawing on signaling and bet-side theory, Colwell et al. (2011) survey 158 purchasing managers from three Canadian manufacturing firms in order to find out how suppliers' code adoption influences buyers' continuation behavior. The results indicate that there is a positive relation between suppliers enforcing CCEs and buyers' commitment to continue sourcing. However, when costs for switching supplier are high, the relation is weakened, indicating that switching costs, as predicted, are a mediating factor. The study provides empirical evidence in a different setting, shedding light on new areas such as incentives of suppliers to enforce own codes. Ciliberti et al. (2011) conduct case studies of four small and medium-sized Italian and Dutch companies that outsource production to examine the effect of codes in the supply chain from an agency theory perspective. SA8000 principles were used as a proxy for codes of conduct. In addition to the visits of the certifying agency, suppliers prepared selfassessment reports and other information as evidence that they were complying with the standards. The results indicate that CCEs can improve communication in the supply chain, and consequently, reduce information asymmetry as well as solve the adverse selection problem in selecting new suppliers and negotiating contracts with existing ones. Furthermore, codes of conduct can solve moral hazard problems as monitoring improves communication, especially for intangible aspect of business, and improves the principal's relation to its suppliers. As such, the findings appear to be heavily dependent on monitoring to reduce information asymmetry.

Effects on Ethical Output (Individual Level)
The second subcategory consists of studies on the effect of CCEs on ethical output (ethical sensitivity, ethical commitment, ethical decision-making, ethical climate etc.). Most of these studies indicate a positive or conditioned positive effect of CCEs and its antecedents, although there is a great mix of measured dependent variables. In contrast to the first subcategory, the empirical contexts encompass North American and Western European companies. Surveys are common but also some case studies are performed. Vitell and Hidalgo (2006) examine, using a survey, Spanish and U.S. employees' perceptions of the role of ethics in organizational success. The tested variables include ethical values, enforcement of CCEs, organizational commitment, ethical idealism and relativism as well as country of residence. Results show that all tested variables, except relativism are positively related to employees' perceptions of the role of ethics in organizational success. Furthermore, the perceptions regarding the role of ethical values were significantly higher among U.S. employees than in Spanish employees. Arnold et al. (2007) survey 297 professional accountants (response rate 54%) in eight Western European countries. The focus was on ethical perception of ethics-related scenarios in the accounting industry. Findings from the study reveal differences across the Western European countries in perceptions of the ethical appropriateness of activities typically covered in CCEs. These differences in perceptions across countries are associated with Hofstede's cultural constructs of Masculinity and Individuality. The findings suggest that a single company-wide code applied internationally will be interpreted and applied differently, even if the company only operates in one geographic region. Tjosvold et al. (2009) use critical incident interviews, asking 101 Chinese employees to describe a situation where their personal ethical values were at issue. Structural equation modeling is used for the analysis. The findings indicate that formal ethics rules in the form of CCEs helped foster a constructive controversy discussion when individuals had opposing views, thus strengthening interpersonal relationships. However, the findings also indicate that CCEs, when only imposed top-down, did not support employees to develop and apply their own ethical values. Petersen and Krings (2009) perform an experimental study of 80 German and foreign managers in Germany. The aim was to study decision-making and personnel evaluations in situations when the CCE was in conflict with supervisors' advice. The findings indicate that if the employee knew the supervisor's position, it affected the decision-making. As such, managers need to 'walk the talk', otherwise, CCEs become "toothless tigers" (p. 510). The authors conclude that "[o]nly when codes are actively enforced and integrated in organizational culture and practice do they re-gain their teeth and become a powerful tool" (p. 510).
McKinney and Moore (2008) mailed a survey questionnaire to a random sample of 10,000 business leaders in the U.S. and report a survey response rate of 12%. Participants were provided a vignette describing international bribery, per the Foreign Corrupt Practices Act, and asked to rate the behavior. The findings show that respondents who worked for firms that have a code of ethics were significantly less likely to rate bribery as acceptable. In addition, the responses suggest that businesses with international revenues were significantly more likely to have adopted a code of ethics than businesses with purely domestic revenues.
McKinney et al. (2010) conducted a survey of 10,000 business leaders in the U.S. in 1993 and again in 2001 (response rate 18% and 12%, respectively) to examine the effects of CCEs on ethical perceptions of actions toward stakeholders. The survey included 16 vignettes depicting scenarios of ethically questionable behavior. The presence of a code had a strong effect on responses to vignettes involving government entities, shareholders, other employees, competitors, and consumers; specifically, the vignettes related to padding expenses, accounting manipulations, underreporting income tax, insider stock purchase, discriminatory hiring practice, violation of copyright, and deceptive advertising. The results indicate that business professionals who work for a company that has a CCE are significantly less accepting of questionable unethical behavior than business professionals who work for a company with no code.
Von der Embse et al. (2010) administered a survey to a national management association in the U.S. (23 useable responses) with mid-and first-level managers from six organizations representing five different industries. The study examines "whether the application of ethical principles in organizational activities is affected by certain ethical safeguards" (p. 4). Safeguards are defined as deliberate measures and programs used to promote ethical practices and prevent unethical behavior. Seven such safeguards are identified (CCE, value statement, written general ethics policies, written specific ethics policies, strong ethical culture, ethics training and readily accessible ethical guidelines). The safeguards are tested in relation to different organizational activities. The results indicate that the most significant safeguards across all areas are CCE, value statement, specific written ethics policies, and readily accessible ethical guidelines.. The least significant safeguard was written general ethics policies. Ruiz et al. (2015) conducted a survey of 525 employees in the finance services sector in Spain (response rate 11.7%) regarding three ethics program components (CCE, ethics training and ethics-oriented performance appraisal content) to find out the relationship to ethical intent. Each component was found to be positively related to ethical intent. In line with expectations, those who perceived all programs to be well implemented scored higher on ethical intent than those with a weak perception of how the programs were implemented. A combination of training initiatives together with one of the other components was equally as effective as all of components together. However, an implementation using only the CCE was effective when training was not applied. Rottig and Heischmidt (2007) administered a questionnaire to MBA students in the U.S. (1998( ) and Germany (2004 to study the impact of ethical training and CCEs in improving ethical decision-making. U.S. corporations are more likely to have a code of ethics and are more likely to provide ethics training than German corporations. U.S. respondents were found, on average, to score higher in ethical decision-making, which suggests the use of code of ethics and ethical training positively affects ethical decisionmaking. The results also suggest that age and years of management experience are positively and significantly related to ethical decision-making. Singh (2011a) conducted a survey of the 500 top ranked Canadian companies (useable response rate 20.9%) to test manager's perceptions of code effectiveness (dependent variable) against various elements of ethics programs (CCE related variables). The result indicates that 58.5% of the variance in the dependent variable (effectiveness of the code) could be explained by the independent variables. A factor analysis of the eighteen items yielded five factors: code purpose, code implementation, internal code communication, currency and external code communication and recency of code utility. The analysis of the relationship between these five factors and the dependent variable indicates that 43.8% of variance is explained by the five factors.
From a reception theory perspective, Norberg (2009) conducted one-on-one interviews with eighteen professional security traders employed by brokerage firms, merchant banks, or mutual funds in Sweden to uncover the brokers and traders perceptions of CCEs. Findings indicate that brokers and traders believe in the invisible hand of markets and prefer not to be restricted by codes and regulations. Regulations and codes are accepted when traders and brokers perceive the restrictions to be supportive of the market. Brokers' and traders' positive attitudes towards codes are pragmatic, e.g., when codes relieve them from personal moral responsibility. When asked about personal morality and the morality of the firm, brokers and traders were found to refer to the code much like the letter of the law. Frostenson et al. (2012) conducted two case studies in retail companies in Sweden involving 33 interviews. The study focused on the internal significance of CCEs in each company-how people refer to the code, whether they find it relevant, and if it is alive in discussions. In both case studies, the CCEs were found to be clearly decoupled. The CCEs did not concern the immediate work context of the employees, but even though the CCEs were not directly relevant for the employees, they were accepted and embraced regarding content, focus and function. However, the CCEs did confirm and strengthen employee identity. The authors pinpoint that "what might appear as a decoupled code cannot be dismissed as irrelevant to the 'core' business processes of the organisation" (p. 263). Valentine and Johnson (2005) administered a survey of 145 respondents from public and private industries in the U.S. Results from the survey suggest a positive relationship between the presence of CCEs and "individuals' beliefs that incorruptibility is an important individual virtue" (p. 45). The authors argue CCEs are useful instruments for organizational change as the codes serve as a base "to the development of employee traits and attitudes instrumental to the advancement of a corporate ethical context" (p. 49). Kaptein (2011) conducts a survey of individuals working for U.S. organizations that employ at least 200 people. The survey was conducted through the National Family Opinion panel database company. 3075 people completed the survey giving a response rate of 77.7%. The study examines the impact of five factors (frequency of communication activities surrounding the CCE, quality of these communications, content of the CCE, embedment of CCEs in the organization by senior and by local management) on unethical behavior. All independent variables were negatively related to observed unethical behavior. The existence of a code has the lowest correlation with unethical behavior, whereas the embeddedness of the code by local management has the highest.
In a related study, Kaptein (2015) conducted a survey through the same database company to examine the effect of nine ethics program components, including the CCE, on unethical behavior. Analysis of the 5065 completed questionnaires reveal that organizations with an ethics program have significantly lower unethical behavior than organizations that do not have an ethics program. The results also indicate that a code of ethics is not sufficient, in and of itself, to reduce the frequency of unethical behavior. In addition, the study finds that the broader the scope of the ethics program, the lower the level of unethical behavior. Further analysis suggests a best sequence for implementing the ethics program components. Five components (CCE, training and communication, monitoring and auditing, accountability policies, and investigation and corrective policies) were found to directly impact unethical behavior. Three other components, ethics officer, ethics report line, incentive and rewards policies, have an indirect impact on unethical behavior. The ninth component, pre-employment screening, was not found to be related to unethical behavior.

Effects on Other Output Variables (Organization Level)
Several studies investigate the effect of CCE on output at the organization level, e.g., CSR, sales volume, and employee turnover. Surveys dominate as the method used, and the use of explicit theory is rare. The overall picture gleaned from these studies is fragmented due to the broad spectrum of research topics covered.
From a meta-level, Graafland and Ven (2011) compare and contrast CCEs in multinational banks in relation to actual behavior in the banking sector. Using MacIntyre's virtue ethics perspective to analyze the bank's CCEs, three virtues are identified which would be needed to fulfill the banking industry's mission: due care, honesty, and accuracy. The underlying question in the study is whether it is expectable that bankers behave differently to these virtues because they work in a context where profit maximization and a shareholder perspective is the only norm. Findings indicate that banks, as well as the individual professionals working in them, can be blamed for actions leading to the financial crisis of 2008, but that the institutional context of the free market economy should be acknowledged. The authors pinpoint a controversy between the incentives in a neoliberal free market system and the core virtues in the financial sector. Institutional changes are argued to be paramount for the financial sector to succeed. Halter et al. (2009) investigate the supplier-buyer relationship through a survey of 30 suppliers to a MNC located in Brazil, to develop knowledge about suppliers' perceptions of the ethical behaviour of staff at purchasing department in Brazil. Data from documented ethical standards, as well as perceptions trust, transparency, respect and loyalty were collected. The results suggest that transparency, supplier awareness of buyer's CCE, and compliance with the CCE, influence the buyer-supplier relationship. The authors further suggest that because of the increased transparency brought about by CCEs, the CCEs act as a tool to reduce corruption. Lee et al. (2014), conduct a survey of 330 service employees (useable response rate 87.9%) at 12 hotels located in South Korea. The findings indicate that corporate philanthropy plays a full mediating role between CCEs and job engagement. Employee awareness of their organization's institutionalization of ethics is directly related to employee awareness of corporate philanthropy. Awareness of corporate philanthropy, in turn, positively influences employee engagement in their job and connectivity to the corporation, which reduces employee turnover intentions.
Some studies connect CCE to CSR (or similar) practices. Erwin (2011), for example studies the effects of code content quality on ethical performance. Using a benchmarking technique, CCEs from 392 international companies and 15 industries were graded on 8 areas of CCE quality. The grades were then used to measure the effects on CSR performance through listings such as Dow Jones Sustainability Index, 100 best corporate citizens, world's most ethical companies and world's most respected companies. Generally, companies with high benchmark scores had higher CSR rankings as well as ethical charts. However, the study also finds that a low quality CCE does not necessitate low CSR performance. The authors conclude that "code quality may play a crucial role in the effectiveness of codes of conduct and their ability to transform organizational cultures" (p. 535).
Mijatovic and Stokic (2010) administered a survey on 852 successful companies in Serbia (14.4% usable response rate) on the influence of internal and external codes on CSR practice. Results indicate that "that internal self-regulations have more influence on CSR practice than the implementation of generic management system standards. However, the existence of transparent corporate values, codes of conduct and implemented management systems according to ISO 9001 and ISO 14001 standards does not prove to be strong predictors of CSR performance" (p. 533).
Using a survey via telephone, interviewing 286 (25% usable responses) procurement managers in German MNCs, Blome and Paulraj (2013) study the effect of CCEs and ethical climate, on purchasing social responsibility (PSR). Findings from the study reveal that top management's ethical stance has an important role in fostering an ethical PSR climate, while CCE implementation has a partial impact. As such, the authors argue for the need of both top management's ethical stance and CCE implementation.
Other studies relate CCEs to aspects that do not have a distinct ethical focus, such as sales volume or customer trust. Gallego et al. (2016), empirically tests the relationship between CCEs, as an illustrator of corporate image and satisfaction, and the effects on sales volume in B2C organizations. Data were collected from 127 Spanish companies with a CCE. The findings indicate that the CCE has a positive impact on sales volume. The author suggests that the code signals increased trustworthiness to the potential customer and improves the company's image as well as prestige, quality and security. Marchoo et al. (2014), conduct an on-line survey distributed to 5000 adults residing in Australia (response rate 18%) to examine travelers' responses to tourism accreditation. Findings indicate that the inclusion of CSR initiatives (accreditation ecolabel and/or a code of ethics) in a tourism brochure increased customer's trust of the tour company, perceived tour value and tour booking intention. The findings also suggest that the inclusion a code of ethics has a stronger effect than the accreditation logo.

Transformation-Oriented Studies
In a previous review, Steven's (1994) noted a lack of studies focusing on how CCEs are communicated, received, and accepted. Helin and Sandström (2007) subsequently reviewed the prevalence and content of such papers between 1994 and mid-2005, concluding that such studies were few, and tended to focus on what to overcome when implementing a CCE. The handful of studies reviewed by Helin and Sandström were conducted in North America and Australia. The papers in this review take on a broader geographical and organizational span, covering different levels of organizational hierarchy. Despite, and perhaps due to, the focus on differing contexts, these studies have begun to shed light on what happens when CCEs are put into practice, in particular by reporting unintended and paradoxical aspects of CCEs.
Approximately, 15% of the total papers reviewed fall within this category. These reviewed papers tend to focus on interpersonal, organizational, legal or cultural tensions in the implementation of codes in different countries or contexts, and are accordingly presented in subcategories which highlight these differences in theoretical or contextual focus. In comparison to Helin and Sandström (2007) where approximately 8% of the papers were classified as transformation and an additional 8% as output/transformation, the relative focus of papers in this category remains similar. Table 3 provides a list of the papers reviewed under the transformationoriented category including author(s) and publication date, subcategory, theory, method, data collection date, context and level of analysis, antecedents, and consequences. This is followed by an overall summary of the papers reviewed within this category, and an individual review of each of the studies, by subcategory.

Summary of Transformation-Oriented Studies
This category reviews 15 studies. With the exception of three studies focusing on legal differences, one longitudinal study applying life cycle analysis (LCA), and one study analyzing pedagogical case responses by Chinese managers, the ten remaining papers utilize ethnographic style case study methods to examine code implementation in different contexts or countries. What differentiates these studies from the content and output studies is that the researchers largely immersed themselves into the studied contexts, bringing to light different aspects of code implementation in practice.
The studies shed light particularly on how other contexts differ from the organization's home country context. On an aggregate level, the transformation-oriented studies thus deal with tensions caused by differences in interpretative (Helin and Babri 2015;Helin et al. 2011;Helin and Sandström 2010;Jensen et al. 2015), legal (Pagnattaro and Peirce 2007;Talaulicar 2009a, b), or cultural approaches (Hanson and Rothlin 2010;Helin and Sandström 2008;Hoivik 2007) in different settings but with the same CCE governance. In contrast to Helin and Sandström (2007) where the focus was mostly on what to overcome when implementing CCEs, these studies attend to where, how and why tensions arise in CCE implementation. However, sometimes specific suggestions for addressing these issues are also discussed.
A number of papers overlap with the subcategories "supplier codes" and "effects on labor conditions" in the contentand output-oriented categories, respectively. These papers deal with the implementation of codes in supplier factories and shed further light on the specific power relations, institutional, historical and cultural aspects associated with code implementation in different supplier factories/countries (Amengual 2010, Bezuidenhout andJeppesen 2011;Raj-Reichert 2013;Rodríguez-Garavito 2005).
In terms of theory, the studies subcategorized as legal are not explicitly theory-driven; yet they refer to national laws, are guided by different ethics principles as well as different approaches to privacy norms. Reference is also made to Hofstede's culture theory. The studies subcategorized as interpretative share a commonality in that they explicitly build on the travel of ideas, and translation literature, which both draw on the works of actor-network theory (e.g. Latour 1987;Law and Mol 2001). The studies subcategorized as cultural vary from a-theoretical, referring to theory on cultural differences (e.g., Hofstede 1991) and building on storytelling and narrative approaches as well as mediation/travel  (Sahlin-Andersson and Sevón 2003;Czarniawska and Sevón 1996) coupled with empirical openness regarding cultural differences.
In terms of data-collection date, the papers that specify the data-collection and field visit dates indicate post SOX empirical observations, however the empirical materials examined date from as early as 1999 through to 2011. The empirical material, hence, is not very recent. The data collection dates could, perhaps be explained by the nature of in-depth qualitative studies, requiring a significant amount of time for write-up, analysis, and publication.
Level of analysis in many of the studies has an (intra) organizational focus, while some focus explicitly on the supply chain, but also incorporate differences in national culture or governance Contextually, the studies focusing on the (intra)organizational level have a broad span with findings stemming from an analysis between U.S. and Western European legal settings (Pagnattaro and Peirce 2007;Talaulicar 2009a, b), between Western Corporations and Chinese subsidiaries (Hanson and Rothlin 2010;Hoivik 2007), Swedish subsidiary of a European MNC (Helin et al. 2011), Swedish subsidiary of a U.S. Corporation (Helin and Sandström 2010).The studies focusing more explicitly on the supply chain report from the Dominican Republic (Amengual 2010), South Africa, Lesotho and Swaziland (Bezuidenhout and Jeppesen 2011), Malaysia (Raj-Reichert 2013), and Mexico and Guatemala (Rodríguez-Garavito 2005).
The transformation-oriented studies differ from the output studies in that they do not have clear antecedents and consequences. As the methods used are qualitative and exploratory the conclusions do not aim to generalize, but rather contribute towards a contextualization of CCEs and how they lead to different outcomes in different situations. The exceptions are the transformation/output studies and the life-cycle analysis by Sethi et al. (2011).
Amassed, we learn from these studies that code implementation involves many different actors and cannot be expected to be a straightforward, copy-paste operation between countries or contexts, but rather demands meticulous negotiations and an understanding of local values, laws, and cultures. We also learn that codes, when implemented in different contexts, may result in a myriad of unintended effects. Codes drafted in one context may clash with deeprooted values and laws in another (Pagnattaro and Peirce 2007;Talaulicar 2009a, b). Codes might also be counteractive to moral empowerment (Helin et al. 2011;Rodríguez-Garavito 2005), simultaneously perceived as good and bad (Jensen et al. 2015) and may be used as tools for domination (Helin et al. 2011). In contexts of competing state and private regulation with limited union impact, CCEs alone may be insufficient in improving workers' rights (Amengual 2010; Bezuidenhout and Jeppesen 2011). Furthermore, monitoring, auditing, and self-regulatory techniques for measuring working conditions may render certain aspects of health and safety visible, while making others' invisible (Raj-Reichert 2013). On the contrary, in contexts where state trust and enforcement is unreliable, a supplementary CCE monitoring system with a clear focus on worker empowerment is argued to provide sustainable worker empowerment and enforcement of workers' rights (Rodríguez-Garavito 2005).

Review of Transformation-Oriented Papers
Legal With a focus on the challenges for U.S. companies operating in the E.U., Pagnattaro and Peirce (2007) add to the specifics of legal clashes by examining the details of conflicts arising when transnational corporations attempt to comply with both U.S. and E.U. laws. With a thorough review of the background and context of laws and provisions regulating CCEs in the U.S. as well as data privacy and worker protection laws in Europe, the authors establish where and when U.S. CCEs are at risk of being in conflict with several laws in the E.U.. Four tried cases and court rulings where a U.S. code of ethics and E.U. legislation have been in conflict are analyzed. The article concludes with specific recommendations in terms of what U.S. firms operating in the E.U. should be aware of, specifically in terms of conflicting historical ideas behind U.S. and European dataprotection and privacy laws.
Elaborating further on the issue of legal clashes, with a theorization of the differences in universalistic and relativistic codifications of ethics, Talaulicar (2009a, b) qualitatively analyze two legal cases dealing with the implementation of U.S. CCEs in Germany. After discussing the legal disputes and proceedings of Walmart and Honeywell's CCEs in Germany, Talaulicar analyzes the contents of the code-finding that both codes are rules-based rather than principles-based, i.e. rather concrete and specific in terms of obligations which employees must observe. The author suggests that a principles-based approach to norm codification might be better suited for global CCEs, because this would allow greater flexibility to respect global ethics while also taking into account local idiosyncrasies.
Cultural Further exploring potential tensions between home and host country in terms of code implementation, and with a focus on differences in national culture, Hanson and Rothlin (2010) provide insight into challenges for Western companies operating in China though the use of autoethnographic data in terms of their own experience working as consultants for Western companies operating, and applying their code of ethics, in China. Based on their experiences over the last 20 years, the authors explain the Chinese legal and organizational context, and suggest that for Western firms to successfully implement their code in China they need to: "Inculturate [the] Code;" "Make the Company Code Consistent with Chinese Laws;" "Align [the] Code with Chinese Concepts and Slogans of Key Government Officials;" "Incorporate References to Global Standards Embraced by the Chinese;" "Publish the Code in Bilingual Format;" "Introduce the Code in the Chinese Way;" "Do Whistle-blowing the Chinese Way"; "Extending the Code to Business Partners;" and "Preparing Local Leadership to enforce the code in China." (pp. 76-78).
On the same theme, Hoivik (2007) draws on responses to 240 case studies assigned to Chinese MBA students who are also managers with several years of experience working for MNCs in China. Hoivik explores how culture influences Chinese managers' perceptions of codes of conduct and performance evaluations implemented by Western corporations, since most Western companies translate their own CCEs literally, into Chinese. Subsequent discussions with other Chinese managers were used to verify the author's understanding of the contents of the case responses. The author concludes that Western codes of conduct are written from a universalist/individualist perspective and hence do not translate to the Chinese culture of particularism/collectivism. Suggestions for Western multinationals include: setting up advisory groups with Chinese experts and employees to evaluate and modify CCEs to fit Chinese norms, annual advisory group meeting to review processes, providing explanations for the underlying values in the code, and regular employee training involving real cases from a Chinese perspective.
Exploring the case of a U.S. parent company code implemented at a Swedish subsidiary, Helin and Sandström (2008) show how the cross-cultural implementation triggered narratives of national belonging and difference among the employees at the Swedish subsidiary. Forty-eight interviews were conducted with nineteen people, using a storytelling approach and qualitative case study method. The main question explored in the study is "Do, and if so, how do, the receivers of the code rely upon cross-cultural aspects when trying to explain the code?" (2008, p. 287). Nine different types of stories relating to cross-cultural differences are drawn upon to show a narrative of belonging and difference which is summarized by the authors as: "Swedish and U.S. societies work very differently. In Sweden, we trust each other and we do not need to explicitly formulate what is common sense the way they have to in the U.S." (p. 287).
Building on the same empirical material as above, Helin and Sandström (2010) further explore the process of implementing a U.S. parent company CCE at a Swedish subsidiary. During the implementation process, the parent company's CCE was forced on the subsidiary without compromise. The study builds on multiple interviews with 19 individuals (a representative sample across departments and hierarchy) of the company's 500 personnel in early-2005 and mid-2006 resulting in a total of 48 interviews. They also studied the CCE, official company materials, company's homepage, company's internal magazine, and ethics and compliance training material. The empirical material is coded using three "editing" processes (Sahlin-Andersson 1996) which highlight how representatives of the Swedish subsidiary explain the American code to themselves and each other to distance themselves but simultaneously manage the implementation: recontextualization (motives), relabeling (package and label), and logic (grouped into five socially embedded rationales or themes). The authors conclude that language structures used in CCE implementation can disempower employees rather than liberating ethical resources.
Interpretative Helin et al. (2011) use a case study method and fieldwork at a Swedish subsidiary to a European Global Corporation. Interviews were conducted at several points in time with HR managers, compliance officers, middle managers, sales employees, and the CSR manager. During the fieldwork, the code became more salient as the researchers were made aware of a recent dismissal of two employees for violation of the code. The case revolves around this incident, and narratives from the field are divided into episodes showcasing how power-relations changed in the lead up to, during and after the dismissal of the two employees. The study theoretically conceptualizes CCEs in terms of control and domination and shows how codes rather than being used as enabling instruments for enlightenment, instead can be used as instruments for domination.
Further exploring CCEs in practice in different cultural contexts, and how these differ, Jensen et al. (2015) use multi-site case studies (Sweden, India, Brazil and Canada) to empirically follow the code and show how enactments of the same code are influenced by different contextual settings. The study conceptualizes the CCE as a fluid object based on object and network theories. The study finds that the significance of the CCE increases farther away from home as the CCE comes closer to encompassing core operations of the business. However, not only is the significance of the CCE increased, but also the code's fluidity; it is enacted in ways that are more abstract and ambivalent. The study concludes by highlighting the complexities in the multiplicity of CCE influence, e.g., codes are perceived to improve the ethical climate and simultaneously lead to concerns about unethical behavior, codes are perceived to improve business and simultaneously be a barrier to business, codes are perceived to increase management control and simultaneously be about common sense, and finally the code is both present and absent.
On a similar note, Helin and Babri (2015) follow the practices associated with implementing a CCE in Sweden as well as practices of monitoring and controlling the implementation of the same CCE in the company's Chinese subsidiary and supply chain. Building on the "travel of management ideas" literature (cf. Czarniawska and Sevón 1996;Sahlin-Andersson 1996), the authors' conclude that the code implemented in practice in the global context entails a negotiation of the ethics in the original document which is transformed through a series of five translations. These five translations are: (i) the code is edited into a re-contextualized signpost from the top; (ii) the code is integrated and edited into a symbol for efficiency and quality; (iii) the code is enacted as a ritual agenda for exercising reversed power;(iv) the code is enacted as a mediating object and contextually bargained; and (v) the code is enacted as a contractual agreement.
Life Cycle Analysis (LCA) In a unique process study of a CCE, Sethi et al. (2011) perform a life-cycle analysis of Mattel Inc.'s code and compliance system called the 'Global Manufacturing Practices'. The case study covers a period of 9 years during which the system was drafted, implemented, and abandoned. Evaluation of both the achievements and the shortfalls in the company's CCE compliance resulted in a bell shaped life-cycle curve. The paper provides a detailed account of the code's life-cycle from the antecedents to the drafting of the code, to the implementation of thoughts into action, through to the code in practice (e.g., training of employees and audits conducted in different countries). The analysis of the case is discussed in terms of the importance of a bold CEO along with support from the board and the stakeholders of the company. The ultimate abandonment of the code resulted from a belief that it no longer served an economic benefit to the company, in addition to the failure of other toy manufacturers to implement similar ethical codes and programs.

Supply Chain
Focusing on the context of supplier factories producing for the apparel industry, and exploring labor conditions, Bezuidenhout and Jeppesen (2011) conduct field research in South Africa, Lesotho and Swaziland. More than 160 interviews were carried out outside 12 factories with workers and informants from governmental agencies, consultancies and trade associations. The focus was on how labor codes play out in practice. A majority of the workers were aware of the code but, in general, workers expressed confusion about whether codes applied to their factories and how inspections worked. The respondents described a lack of enforcement of rights and standards from government as well as pressure for workers to put on an act during inspections to retain contracts with buyers. Unions were found to be in a difficult position and only able to make limited impact and did not use codes as tools for organizing. The conclusion of the study is that, in the absence of coherent and global governance structures for a particular industry; codes will not have any material impact on workers' rights in southern Africa or globally.
In addition, focusing on the apparel industry, in the Dominican Republic, where diverse regulatory practices converge, Amengual (2010) studies two organizations, one multinational organization and one labor organization, to explore the core question, "What happens when state and private regulation meet on the factory floor?" (2010, p. 405). The author conducted over 100 interviews complemented with document analysis and observation to research the different kinds of state and private regulation and to explore how these regulations interact on the factory floor. The comparison of, and tensions between, state and private regulation are explored against a backdrop of debates concerning the effectiveness of private regulatory systems, i.e., CCEs and concerns that these may counteract or displace state regulation. In the case of the Dominican Republic, Amengual (2010) does not find evidence for state displacement by private regulation nor does he find evidence for active state and private coordination on the factory floor. Amengual (2010) argues, that both systems have comparative strengths and advantages in terms of observing violations, sanctioning firms, and sharing knowledge, and, as such, improving labor conditions is dependent upon leveraging comparative advantages from both, complementary systems.
Focusing on the role of health and safety of workers in the safeguarding of labor rights, Raj-Reichert (2013) explores the different practices and institutions that regulate and affect the safety and health of workers of Malaysian manufacturers in the Hewlett Packard supply chain. Raj-Reichert's case study is based on fieldwork and interviews during 2009 and 2010 at printed circuit board manufacturing sites in Malaysia. Interviews are conducted throughout the supply chain, with government officials, global and local civil society organizations, and union representatives. The study finds that the Malaysian government transferred health and safety responsibilities to the industry employers, which in turn employed safety and health officials (SHOs) who did not have in-depth knowledge of the health risks associated with the work environments. Interviews reveal that the SHOs believed the ultimate benefit of compliance was to showcase their results to make the customer happy and thereby maintain profits. Factory workers' health and safety training was metrics driven, but the workers were given limited information about health and safety and were not informed of applicable laws. Based on the results of the study, the author argues that SHOs make management from a distance possible, yet "there is essentially a delinking of worker health impacts and conditions in the workplace from the goals and objectives of the governance systems undertaken by firms. Thus, while self-regulatory techniques such as standards, benchmarks and audits render certain aspects of the health and safety domain visible, others such as worker health conditions, particularly in the long run, fall in their 'shadows'" (2013, p. 28).
Based on multi-sited ethnographic research in the apparel industry in Guatemala and Mexico, Rodríguez-Garavito (2005) explores the role of code related monitoring on workers empowerment and working conditions. Rodríguez-Garavito (2005) interviews a diverse range of people including factory owners, state regulators, legal counsels, and leaders of unions and federations in addition to factory visits. Basing the effect of monitoring systems on potential sources of governance failure, the author proposes a model called the "empowered participatory labor regulations (EPLR)" (p. 211) which would supplement rather than replace state regulation. With a close eye on the governance structures (state, corporate, union), the author suggests that "different types of codes and monitoring have different effects on the enforcement of labor rights" (p. 206). The study proposes that, in environments where state trust and law enforcement is unreliable in terms of protection of workers' rights, a private system for monitoring together with a clear focus on worker empowerment, could be beneficial for the sustainable enforcement of workers' rights in global factories.

Discussion and Conclusions
The purpose of this study was to provide an updated answer to the question identified by Stevens (1994) and approached by Helin and Sandström (2007), now asking a further decade later, 'where are we now?' in terms of empirical CCE research. To answer this question, the empirical literature has been reviewed focusing on empirical foci, context, questions addressed, methods, findings and theory. The remainder of this paper is dedicated to providing an overview of the findings from the review, discussing whether the gaps related to how codes work has been adequately filled, and discussing potential avenues for future CCE research.
First, the review finds that CCEs have continued to converge and use and authoritarian language of control. From studies focusing on content, and what is in the actual code, we learn that regulatory and legal influence on code content is a consistent theme. Holder-Webb and Cohen (2012) believe this has led institutional isomorphic convergence in code content. Other studies also find convergence of code content (Forster et al. 2009;Sharbatoghlie et al. 2013). In addition, studies of codes for companies that operate internationally find that codes do not address, and lack flexibility/adaptability in code content, to local laws, values and norms (Amaeshi and Amao 2009;Halff 2010;Svensson and Wood 2007). Examinations of the language (symbolic aspects) of code content find that codes reinforce corporate hierarchy with a top-down, authoritarian emphasis (Winkler 2011;2012;Munter 2013) and this is found also in a case study of a CCE in practice where Helin et al. (2011) find that codes rather than being used as enabling instruments for enlightenment, instead can be used as instruments for domination. Forster et al. (2009) also find imitative features for codes, especially for small companies, which copy large company code content. The content of MNC supplier codes appear to be a response to public criticism, anticipated legislation or competitive advantage, but pass responsibility for ethical, environmental and social issues to the supplier while ignoring the supplier's economic issues (Preuss 2009(Preuss , 2010. Second, the review finds that CCEs are perceived to be positive or neutral, but actually lead to both positive and negative outcomes. As suggested by Kaptein and Schwartz (2008) a distinction between actual and perceptual effectiveness in this review shows a stark bias towards studies utilizing perceptual data. Regarding the perceptual effects, differences are found in perceptual effects in different countries (Arnold et al. 2007;Vitell and Hidalgo 2006) and between companies with or without a CCE (McKinney et al. 2010;McKinney and Moore 2008). CCEs are positively related to individual' beliefs that incorruptibility is an important virtue (Valentine and Johnson 2005). Ethical intent is positively related to CCEs, ethical training and ethical performance appraisal content (Ruiz et al. 2015). Furthermore, the idea of profit maximization that prevails in the finance industry is found to contrast with the underlying virtue that CCEs might create (Graafland and Ven 2011;Norberg 2009). Corporate philanthropy has a mediating role between CCE and job engagement (Lee et al. 2014). In addition, codes may be accepted and embraced even if they are decoupled from core business (Frostenson et al. 2012). Suppliers' perception of buyer's compliance with CCE through transparency, respect and loyalty is positively associated with the buyer-supplier relationship (Halter et al. 2009). In addition, active enforcement and integration are perceived as necessary for CCEs to be effective (Petersen and Krings 2009).
In terms of actual effects, the few existing studies indicate that CCEs can foster discussion when individuals have opposing views, but do not empower own ethical values (Tjosvold et al. 2009). Ethical decision-making is affected by years of management experience and training (Rottig and Heischmidt 2007). Observed unethical behavior is negatively associated with CCEs, CCE communication, and the embeddedness of CCEs (Kaptein 2011(Kaptein , 2015. Companies with high ethical quality in their CCE content also have higher CSR rankings (Erwin 2011) but on the other hand Mijatovic and Stokic (2010) indicate that CCEs are not a strong predictor of CSR performance. Top management's ethical stance in fostering an ethical climate positively affects purchasing social responsibility (PSR). CCEs have a positive effect on sales volumes (Gallego et al. 2016) and customer trust (Marchoo et al. 2014).
Third, the review finds that CCEs are not sufficient, but sometimes even counterproductive in terms of improving labor conditions and workers' rights. Studies reporting specifically on effects in the supply chain show both positive and counteractive results, indicating that the effects of the CCE are conditioned by top management commitment (Withers and Ebrahimpour 2013), employee commitment (ibid) embeddedness (Kaptein 2011), scope and composition of ethics programs (Kaptein, 2015) and pressure from consumers (Koçer and Fransen 2009). Other mediating factors include switching costs (Colwell et al. 2011) and monitoring to reduce information asymmetry (Ciliberti et al. 2011). CCEs as a form of organizational regulation together with complementary government regulation is suggested as important for improved workers' rights and conditions (Egels-Zandén 2014). When focusing on the implementation of CCEs in the supply chain, studies highlight structural tensions arising from the power relations made visible between buying and supplying ends of the same governance structure (Amengual 2010;Bezuidenhout and Jeppesen 2011;Raj-Reichert 2013;Rodríguez-Garavito 2005). Such conflicts between the ethics in the code and the working conditions of factory workers may further create incentives for selective auditing and monitoring (Raj-Reichert 2013). Such tensions may also pertain to aspects related to profit maximization goals and the competition on the market (Graafland and Ven 2011;Yu 2008). In conclusion, CCEs on their own seem insufficient in terms of improving workers' rights in factories located in different cultural, geographic and socioeconomic settings.
Fourth, studies focusing on CCEs in different contexts highlight the complexity, inflexibility and dark side of CCEs in practice. Codes of ethics, when implemented as governance systems spanning different national boundaries and organizational hierarchies, encounter several types of tensions. The studies focusing on how CCEs are translated in practice highlight legal (Pagnattaro and Peirce 2007;Talaulicar 2009a, b) as well as cultural (Hanson and Rothlin 2010;Sandström 2008, 2010;Hoivik 2007) tensions between parent company/country and subsidiaries. CCEs emerge from these studies as complex tools, as they are shown to be counteractive to moral empowerment (Helin et al. 2011;Rodríguez-Garavito 2005), simultaneously perceived by organizational members as good and bad (Jensen et al. 2015), and might be used as tools for domination (Helin et al. 2011).
Fifth, similar to Helin and Sandström's (2007) conclusions, the question of how codes work, based on qualitative data based on practice, is still the least addressed question, with questions pertaining to the content of codes and effectiveness of codes still dominating the literature. Several papers are now driven by theory, however the majority of papers are still either not explicitly grounded in theory or completely a-theoretical. There has been an increase in contextual extensiveness, however, with the added effect of dispersion and a lack of integration between studies in terms of definitions of variables, contextual comparison and methodologies. This ultimately leads to a wide spectrum of individually interesting studies in different settings, but a lack of cohesive knowledge generation regarding CCEs.
Taken together, in the stock of empirical research, CCEs emerge as complex tools which companies implement in rather similar tones and with similar content. CCEs are perceived to be effective, and seem to be effective in terms of controlling unethical behavior to a limited extent, but at the same time CCEs may be morally disempowering. The context in which CCEs are put to work and studied, is heterogeneous (cultural, legal, hierarchical, organizational), and there are numerous factors influencing the outcomes of CCEs. CCEs, often formulated by corporate attorneys, are concerned with protecting the company, and may be counteractive in terms of the effects they have in different settings. This is in line with Steven's (2008, p. 607) conclusion indicating that "codes do not work when they are written by management and passed down to employees as a mandate". Given the findings, we argue that governance systems should attend to different groups of stakeholders in different ways, where some may need to be monitored, while others need protection or both. The review also indicates that for CCEs to be effective in a positive way, commitment from both management within the organization, consumers outside the organization, and government regulation for workers' rights is necessary, thus signifying the importance of positive input from several stakeholders.
Based on the review of papers and discussion, we see several areas of viable and relevant exploration in the field of CCEs. (1) Expand on the studies of discursive characteristics of codes using different methodologies not only in content oriented studies but also output and transformation orientations, e.g., interviews with employees, managers, or subcontractors. (2) Similar to the labor focus studies, case studies focusing on the effects of codes on the environment, i.e., from a social focus to encompassing an environmental focus. (3) Explore components of existing frameworks (e.g., Kaptein and Schwartz 2008) to build a consistent body of research to make possible future meta-analyses of the content and output studies. (4) More transformation-oriented studies at different organizational levels and in different geographical contexts.

Limitations
As with all studies, there are limitations to this study. Firstly, the total number of studies reviewed is limited by our selection criteria, where we have chosen to select papers based on specific keywords, in specific search engines. There is always the risk of missing out on some papers, however, we have discussed the search methodology and criteria for selecting papers to be as transparent as possible. Second, the three overarching categories; content, output and transformation are blunt and at times overlapping. We address this issue in the appendix by mentioning overlaps. In addition, we devote some text to review each paper to give the readers an idea of what the paper is about rather than just assigning them to a category. Third, there is a limitation to aggregating findings from studies using different methods and carried out in different contexts as the aggregate discussion necessarily must take on a meta-view and hence may miss important details. We do; however, attend to details in the reviews of the individual papers and the initial tables. Fourth, our selection criteria did not include studies from non-business settings, as such, there may be differences between the conclusions reached from the studies reviewed and those in a public sector context.