Why Do Managers Leave Their Organization? Investigating the Role of Ethical Organizational Culture in Managerial Turnover

The aim of the present longitudinal study was to quantitatively examine whether an ethical organizational culture predicts turnover among managers. To complement the quantitative results, a further important aim was to examine the self-reported reasons behind manager turnover, and the associations of ethical organizational culture with these reasons. The participants were Finnish managers working in technical and commercial fields. Logistic regression analyses indicated that, of the eight virtues investigated, congruency of supervisors, congruency of senior management, discussability, and sanctionability were negatively related to manager turnover. The results also revealed that the turnover group is not homogeneous, and that there are several different reasons for leaving. The reasons given for turnover were grouped into five different categories: (1) lay-off, (2) career challenges, (3) dissatisfaction with the job or organization, (4) organizational change, and (5) decreased well-being/motivation. ANCOVA analyses showed that those managers who stayed in their organization perceived their ethical culture to be stronger than those in turnover groups, and especially compared to groups 3 and 5. The results acquired through different methods complemented and confirmed each other, showing that by nurturing ethical virtues an organization can decrease job changes and encourage managers and supervisors to want to remain in their organization.


Introduction
Employee turnover and its antecedents have been popular topics in research over the decades (see, e.g., Griffeth et al. 2000;Hom and Kinicki 2001), but managers' departure from an organization and especially the reasons leading up to it have been much less studied. There is a gap in research especially in terms of longitudinal studies on actual job change among managers, although the consequences of managers exiting an organization may be particularly significant. The costs of turnover are considerable for organizations (Moynihan and Pandey 2008;Simmons 2008), particularly in terms of search and recruitment as well as company performance (Shaw et al. 2005). Turnover among managers may create uncertainty among employees and their beliefs about the future of the organization (Klotz and Zimmerman 2015), and replacing a manager may be difficult and time-consuming. As one of the most powerful sources of long-term competitive advantage lies in human and social capital (Becker et al. 2001;Pfeffer 1995), the departure of managers may endanger inter-organizational relationships and cause a loss in firm-specific human capital (Furtado and Karan 1990), as well as having negative effects on organizational performance (Hill 2009). Therefore, more research on managerial turnover is needed.
In such scarce research as there is into manager turnover, a few antecedents have been investigated, including organizational commitment (e.g., Carbery et al. 2003;Udo et al. 1997;Rosin and Korabik 1995) and job satisfaction (Carbery et al. 2003;Rosin and Korabik 1995;Vidal et al. 2007) which have mainly been studied in cross-sectional settings and in terms of turnover intention. The role of ethical organizational culture has remained largely unexplored in managers' job change decisions, which is remarkable, since the organization's ethical culture may well be one of the key elements in keeping valuable managers in the organization. Ethical organizational culture supports managers' well-being (Huhtala et al. 2011 and it is associated with less sickness absences (Kangas et al. 2015), more organizational trust (Puč _ etait_ e et al. 2015), and commitment to the organization (Huhtala and Feldt 2016). Unethical culture, on the other hand, has been shown to increase managers' probability of having work goals related to changing jobs or giving up their career (Huhtala et al. 2013). These findings show that there is reason to assume that an ethical organizational culture would play a role in reducing manager turnover.
To verify the nature of the associations and possible causality between ethical culture and manager turnover, longitudinal research is needed to establish ethical culture as a predictor of later job changes. In the present 4-year longitudinal study, the actual turnover among more than 450 Finnish managers in the context of ethical organizational culture will be examined. This is the first study to investigate the self-reported reasons leading to managers' job changes, and the virtues of ethical culture associated with those reasons. The study of reasons managers give for leaving an organization can provide deeper insight into what leads to the decision to exit the organization. By combining quantitative data with openended questions and using a longitudinal design, it is possible to gain a more comprehensive understanding of the phenomena and the causal associations between them, possibly to better comprehend and prevent job changes in organizations. The role of ethical organizational culture as an antecedent of manager turnover may be crucial, since an organization nurturing ethical values can be seen as a favorable working environment, supporting, for example, well-being and commitment to the organization.

Ethical Organizational Culture
An ethical organizational culture is the ethical quality of a work environment, incorporating the experiences, presumptions, and expectations of the organization's members as to how the organization can prevent unethical behavior and encourage ethicality (Kaptein 2009). Ethical culture promotes organizational virtues and consists of the shared values, norms, and beliefs provided to the organization's members (Kaptein 2008;Treviño and Weaver 2003). In the literature, the concepts of ethical culture (Trevino 1990) and ethical climate (e.g., Victor and Cullen 1988) are seen as strongly related, but independent (Kaptein 2008(Kaptein , 2011Treviño et al. 1998;Treviño and Weaver 2003). Whereas ethical climate relates to the atmosphere and conditions in the organization or work unit and defines the perceptions of the organization's members about what is ethical behavior (Heugens et al. 2006;Treviño and Weaver 2003), ethical culture can be seen as a deeper and broader construct that includes the factors that guide compliance with the ethical expectations and enhance ethical behavior in the organization (Kaptein 2011;Trevino 1990).
To date, the most extensively tested model of ethical culture is the CEV model developed by Kaptein (2008). It is based on virtue ethics, which defines the morally right, virtuous way of behaving. Virtue ethics has its roots in the Aristotelian approach, which was incorporated into business ethics by Solomon (1999Solomon ( , 2000Solomon ( , 2004. At the organizational level, virtuousness can be seen from the way in which the culture encourages employees to act ethically and prevents their unethical behavior (Kaptein 1998(Kaptein , 1999. However, as Kaptein (2015) recently noted, corporate virtues lie, in fact, mid-way between two vices: unethical behavior can occur both when the virtue is absent and when it is present to excess. Therefore, organizations should find a balance and avoid either extreme. Virtuousness in organizations has been found to have positive effects, such as creating self-reinforcing positive spirals, and strengthening and protecting organizations from traumas such as downsizing (Cameron et al. 2004).
According to the CEV model, an ethical organizational culture includes eight virtues that promote ethical conduct if their presence in the organization is strong. In the present study, these virtues will be used to represent the ethical culture as a whole: (1) Clarity refers to the extent to which ethical expectations are made concrete and understandable in the organization; (2) Congruency of supervisors and (3) Congruency of senior management represent the degree to which managers act as role models, following the organization's ethical standards; (4) Feasibility refers to the conditions created by the organization to enable employees to fulfill their responsibilities; (5) Supportability refers to the extent to which the organization provides support for management and employees to identify with the ethics of the organization; (6) Transparency refers to the degree to which the consequences of conduct are visible to everyone in the organization; (7) Discussability stands for the opportunities that exist to discuss ethical issues internally; and (8) Sanctionability is defined as the extent to which employees believe that wrongdoing is acknowledged and punished, and ethical behavior rewarded.

Manager Turnover
Classic studies of managerial turnover largely depended on measuring organization-level variables, such as size, organizational demography, and performance (Grusky 1961;Harrison et al. 1988;Wagner et al. 1984), whereas other, ''softer'' aspects of work remained uncharted. Later, some studies have shown that, for example, managers' job satisfaction (Carbery et al. 2003;Rosin and Korabik 1995;Vidal et al. 2007) and organizational commitment (e.g., Carbery et al. 2003;Udo et al. 1997;Rosin and Korabik 1995), which have been the two most studied antecedents of turnover, are associated with fewer turnover intentions. Even though it is known that managerial turnover can have a negative impact on an organization's vitality (Furtado and Karan 1990) and performance (Hill 2009), and despite many calls for additional research on the antecedents of actual turnover among managers (Hambrick et al. 2005;Lee and Ashforth 1993), most research has focused on turnover intentions, and among lower-level employees. Moreover, a clear shortage in research has been the lack of longitudinal settings, since most of the studies examining turnover have in fact been cross-sectional and studied turnover intention.
Studies on the reasons managers themselves have reported for their turnover are scarce. Particularly qualitative research is lacking, except for two studies. The first study that examined reasons for turnover showed that among hotel general managers (n = 52), the most commonly cited reason for turnover was a conflict between the manager and the owner of the property (Birdir 2002). In the second study, which included 238 male and 303 female managers, a 17-item survey on turnover reasons was generated following pilot sample interviews (Rosin and Korabik 1995). Factor analyses revealed that there were two dimensions characterizing turnover: work reasons (e.g., lack of opportunity, long hours, office politics) and family reasons (e.g., work-family conflict), but neither of them were significantly associated with turnover. However, these studies had their limitations: either they were conducted over 20 years ago or had a very small sample size and lacked a theoretical framework. Since then, working life has changed and ethical and moral questions have become more visible and discussed. Especially, the ways in which the organization operates and the set of values it follows have become essential for individuals seeking jobs. Regardless of these developments, more detailed qualitative information on the reasons managers give for their job change remains unclear, and especially in relation to the ethical organizational culture, which represents organization's values.
In examining the reasons for changing or leaving an organization, ''push'' and ''pull'' motivators have been exploited (e.g., DeHoog and Whitaker 1990;Lee and Mitchell 1994;Preenen et al. 2011). In fact, most turnover research has concentrated on ''push'' motivators (e.g., attitudes toward the current organization), whereas less attention has been paid to ''pull'' factors that refer to attraction from outside the organization (e.g., job offers). Push-motivated quitting has been defined as a movement away from one's job as a consequence of an imbalance between the perceived costs and benefits of staying (March and Simon 1958). In a study of turnover among city managers, for instance, push factors were seen to derive from political conflict or differences in style, orientation, or policy between managers and their municipal councils (McCabe et al. 2008). Sometimes, managers themselves can choose to resign when they face pressure; studies have shown that, for instance, breaches in the psychological contract (Carbery et al. 2003;Collins 2010;Shahnawaz and Goswami 2011), possible barriers to career advancement (Rosin and Korabik 1995), and high levels of role conflict (Good et al. 1988) may lead managers to leave their organization even though they would not have a new job available. Pull-motivated quitting, on the other hand, may induce managers to look for new positions not because of dissatisfaction with the current organization, but because attractive financial, personal, or professional career advancement possibilities opened up to them (DeHoog and Whitaker 1990; Mowday et al. 1982).
In the present study, it is argued that ethical organizational culture creates a good basis for attracting and keeping managers in their organizations. It is proposed that push-motivated job changers gave more negative evaluations to the previous organization's ethical culture compared to pull-motivated job changers. Pull-motivated change of jobs is not expected to associate with previous perceptions of the organization's ethical culture. Moreover, the expectation was that those managers who had stayed in one organization during the study period would perceive their organization's ethical culture as stronger than those managers who had changed their jobs because of push motivators. An ethical organizational culture can be seen as a favorable working environment that creates a good basis for attracting and keeping employees and managers in the organization. To explain the process of how an ethical organizational culture might influence manager turnover, the theories of psychological contract and P-O fit can be used. A psychological contract is defined as ''an individual's belief regarding the terms and conditions of a reciprocal exchange agreement between that focal person and another party'' (Robinson and Rousseau 1994, p. 246). The theory is rooted in social exchange theory (Blau 1964). It has been studied mainly from the perspective of employees who have perceived a breaching of the psychological contract or the failure to fulfill it (e.g., Guest 2004;Robinson and Rousseau 1994;Sturges et al. 2005). A violation of the psychological contract has been shown to relate to job dissatisfaction, lowered professional commitment (Suazo et al. 2005), absenteeism (Deery et al. 2006;Johnson and O'Leary-Kelly 2003;Sturges et al. 2005), and turnover intention among both employees (Suazo et al. 2005;Sturges et al. 2005;Carbery et al. 2003) and managers (Shahnawaz and Goswami 2011). Sims (1991) has proposed that organizations can enhance the institutionalization of ethics into their daily conduct by managing psychological contracts and creating and nurturing an ethically oriented organizational culture. However, small but significant differences in expectations between the organization and its employees' psychological contract can cause mismatches in ethical conduct. According to Sims (1991), the clarification of ethical issues in the psychological contract can contribute to building a strong ethical organizational culture, which in turn has been found to relate to fewer turnover intentions (e.g., Pettijohn et al. 2008;Valentine et al. 2011). In addition, ethical culture (Andrews et al. 2011) and ethical values have been shown to actually strengthen the psychological contract by improving social exchanges within an organization (O'Donohue and Nelson 2009;Rousseau 1995;Sims 1991;Valentine et al. 2002). Interviews with managers with turnover intentions have revealed that managers with intent to leave were disappointed at the unfulfilled commitments made in the recruiting phase, while they perceived that they themselves had honored their responsibilities (Collins 2010). Thus, a breach in the psychological contract that took the form of non-transparent and unethical conduct was found to relate to turnover intention. Based on the reasoning above, we assume that an ethical organizational culture can contribute to the fulfillment of the psychological contract and reduce turnover among managers.
Person-organization fit (P-O fit) refers to the relationships individuals and organizations have. P-O fit has many forms, one of which is congruency between the individual's own and the organization's values (Chatman 1989(Chatman , 1991O'Reilly et al. 1991;Kristof 1996). A fit between person and organization increases employees' commitment and reduces turnover intentions (e.g., Cable and Judge 1996;Chatman 1991). The P-O fit can also be extended to, for instance, a person-people fit, the fit between employees and employers, and to the value alignment between colleagues, all of which play a role in turnover decisions (Kristof 1996;Vianen 2000). Coldwell et al. (2008) assume, following P-O fit theory, that individual perceptions of ethical organizational fit depend on their perceptions of the organization's ethics and reputation. They propose that an organization's ethical orientation and reputation impact significantly on the employees' sense of P-O ethical fit, and this in turn affects both their feelings of overall satisfaction with the organization and their retention (Coldwell et al. 2008). A fit between an employee's individual ethical orientation and the organizational ethics reinforces positive attitudes and behaviors in the organization. A few studies have found that a misfit between personal and organizational values creates ethical conflict, which might lead to turnover (Schwepker 1999;Thorne 2010). In a recent study, ethical culture was found to relate to fewer turnover intentions as well as better P-O fit, which in turn partially mediated the connection between ethical culture and employee outcomes (Ruiz-Palomino et al. 2013). Thus, it is probable that managers are more likely to choose to stay in organizations that have an ethical culture than in those that uphold unethical values.

The Aims of the Present Study
The first aim of this study is to quantitatively investigate whether ethical organizational culture predicts managers' job changes. Another aim is to examine what kind of reasons managers have retrospectively given for their first change of jobs, and whether the virtues of an ethical organizational culture are associated with these reasons. We suggest that those managers who have more negative perceptions of their organization's ethical culture will name reasons for changing their job that are related to push motivators, i.e., to conflicts or differences between the managers and their organization. Managers may also give reasons related to pull factors, i.e., reasons related to seeking career development and new challenges, which are less likely to be strongly associated with ethical organizational culture. In this study, the multidimensional CEV model is utilized to study ethical organizational culture. The advantage of this model is that it provides us a more detailed understanding of which virtues of ethical culture are most important in preventing managerial turnover and are related to the reasons for changing of jobs. In addition, since this study includes both open-ended and closed questions about manager turnover, information about the same phenomenon can be gained from different perspectives. Based on the aforementioned reasoning, the following hypotheses were formulated: H 1 An ethical organizational culture predicts manager turnover.
H 2 An organizational culture that is perceived as less ethical is associated with turnover reasons related to organizational ''push'' factors.

Method Data Collection
To gather the data for the present study, we chose a four-year three-wave approach. The data were collected in multiple waves because this allows us to study the changes in variables from one period to another (Menard 1991) as well as connections between events that are separated in time (Hakim 1987). The data were collected in three time points: 2009 (T1), 2011 (T2), and 2013 (T3). The 2-year time frame between the studies was chosen since adequate time is needed for actual turnover to occur, and for the study of its possible antecedents. The original sample of managers working in technical and commercial fields (N = 3000) was randomly selected from the membership registers of two Finnish national labor unions (The Finnish Association of Business School Graduates, n = 1500, and The Finnish Association of Graduate Engineers, n = 1500). According to statistics presented by Ahtiainen (2011), 67 % of employees in Finland in 2009 were part of a labor union organized according to industry, which is why our sample is relatively representative of Finnish managers graduated from business and engineering schools.
In 2009 (T1), the questionnaire was posted to 3000 members of these two unions, of whom 369 returned an uncompleted questionnaire because they were not included in the target group (e.g., did not work in a managerial position, were unemployed, or had retired). Altogether, 902 managers returned a filled-out questionnaire, a response rate of 34 % after removing those 369 empty forms from the original sample (for more detailed background information, see Huhtala et al. 2011). The majority of respondents were men (70 %, n = 629), the biggest single group worked in upper middle management (36 %, n = 327), and the average age of the respondents was 46 years (range 25-68, SD = 9.3).
In 2011 (T2), the questionnaire was sent to those managers who had participated in the study at the baseline measurement and had not declined to be contacted again (N = 728). In total, 491 questionnaires were returned. Of the respondents, 27 returned an empty form (e.g., unemployed, retired) and thus 464 were included in the final sample, yielding a response rate of 64 %. That is, of the baseline sample in 2009 (T1; n = 902), 51 % also participated in the follow-up study in 2011 (T2).
In 2013 (T3), the questionnaire was sent again to those 687 managers who had participated at the study baseline and had not in previous measurements declined to be contacted again. Altogether, 455 respondents participated at this stage of the study, yielding a response rate of 66 % in 2013 (T3).

Participants in the Present Study
In the present study, we investigated managers' perceptions of the ethical culture of their organization using the 58-item Corporate Ethical Virtues (CEV; Kaptein 2008) questionnaire, which is presented in more detail below. Managers answered the CEV questionnaire at two measurement points, 2009 (T1) and 2011 (T2). The turnover of managers was measured in 2011 (T2) and 2013 (T3). The association between ethical organizational culture and turnover was studied in three follow-up settings, the first two of which were two-year periods and the third one a four-year period: (1) T1 (CEV) to T2 (turnover), (2) T2 (CEV) to T3 (turnover), and (3) T1 (CEV) to T3 (turnover). Three different subsamples were formed to examine the ethical organizational culture as a predictor of manager turnover between these three time lags. A fourth subsample was used to study the associations between ethical culture and the reasons for turnover given by the managers. The study setting is illustrated in Fig. 1.
Subsample 1 In the first phase, we included those managers (n = 451) who had evaluated their ethical organizational culture at the study baseline in T1 and provided information in T2 about whether they had changed their job during the previous two years (T1-T2). The majority of the respondents were men (68 %, n = 305), and they were 46 (SD = 9.02) years old, on average. Of the managers, 18 % (n = 80) worked in lower middle management, 36 % (n = 165) in upper middle management, 33 % (n = 148) simply gave their position as in management, and 13 % (n = 58) worked in upper management.
Subsample 2 In the second phase, we included those managers (n = 308) who had filled in the CEV questionnaire at T2, evaluating their current employer at that time, and who had indicated in T3 whether or not they had changed jobs during the previous two years (T2-T3). Most of the respondents were men (68 %, n = 209), with the average age being 48 (SD = 9.10) years. Of the respondents in 2013, 19 % (n = 59) worked in lower middle management, 33 %, (n = 99) in upper middle management, 30 % (n = 90) in management, and 15 % (n = 47) in upper management.
Subsample 3 In the third phase, we included those managers (n = 410) who had filled out the CEV questionnaire in T1 and who reported in T3 whether or not they had changed their job during the previous four years (T1-T3). We wanted to see if the organizational culture evaluated at T1 could predict the first job change during this four-year period. In this sample, most of the respondents were men (68 %, n = 280), with the average age being 50 (SD = 9.02) years. Of the respondents, 16 % (n = 59) worked in lower middle management, 29 %, (n = 111) in upper middle management, 33 % (n = 125) in management, and 16 % (n = 62) in upper management.
Subsample 4 The fourth subsample consisted of those managers who had reported the reasons for their first job change during the T1-T3 time period (n = 124), and those managers who had not changed their job during this time period (n = 258, referred to as the group of stayers). Ethical organizational culture measures were added to the analyses from either T1 or T2, depending on which time point preceded the first change of job and the reasons given by the managers. Of the managers, 68 % (n = 261) were men, they were on average 50 (SD = 8.39) years old (measured in 2009), and 16 % (n = 62) worked in lower middle management, 36 % (n = 138) in upper middle management, 34 % (n = 130) in management, and 14 % (n = 52) in upper management.
The attrition analyses comparing gender, age, and managerial level among the participants were done separately for the four subsamples used in this study. In the attrition analyses, those who belonged to the subsamples (i.e., had provided information about their job changes and about their perceptions of ethical culture) were compared to the ones who had not participated in the study after the baseline measurement in 2009. In Subsample 1, the analysis showed that the respondents at T2 (n = 451) did not differ from those who had left the study after the baseline measurement (n = 451). In Subsample 2, no differences were found between the respondents at T3 (n = 308) and the ones who had left the study (n = 594), and the same applied to Subsample 3 which included 410 respondents, and to Subsample 4 including 382 participants.

Measures
Turnover was investigated in T2 and T3 by asking participants whether they had changed or left their job during the previous two years (yes/no). These variables were used to examine turnover in Subsamples 1, 2, and 3. The turnover information reported by the participants was double checked and compared to the background variables reported two years earlier (e.g., years worked in the company).
An open-ended question at T3 asked about the exact timing and major reason for changing job. A 'turnover calendar' was developed for this study, on which the respondents were asked to mark the year and month they had changed their jobs. They were also asked to write the reason(s) for their move in a separate empty space. The advantages of this type of open-ended question are, for instance, that it allows the respondents to include more information, including feelings and attitudes, to answer in detail, and to qualify and clarify responses. The reasons the managers had retrospectively given for their first change of jobs (between years 2009 and 2013) in the open-ended answers were categorized into five different turnover categories that will be presented later. These categories were used in Subsample 4 to examine the associations between ethical culture's virtues and reasons for changing jobs.
Ethical organizational culture was assessed by the 58-item Corporate Ethical Virtues (CEV) questionnaire at T1 and T2 (Kaptein 2008). The 58-item CEV scale is the only normative and multidimensional scale measuring the Fig. 1 a Theoretical model of the study setting for Subsamples 1, 2, and 3. b Theoretical model of the study setting for Subsample 4. CEV a was used when the first turnover had occurred between T1 and T2; CEV b was used when the first turnover had occurred between T2 and T3 ethical culture of organizations that has been empirically validated (Huhtala et al. 2011;Kangas et al. 2014;Kaptein 2008). The eight-factor structure of the CEV scale has also been found to remain unchanged across groups (Kangas et al. 2014) and across time in the present longitudinal data . The original Dutch version of the scale was first translated into Finnish and then back into Dutch by two independent authorized translators (see, e.g., Huhtala et al. 2011). The scale consists of the following eight subscales measuring each virtue: clarity (10 items, e.g., ''The organization makes it sufficiently clear to me how I should deal with confidential information responsibly''); congruency of supervisors (6 items, e.g., ''My supervisor does as he says''); congruency of senior management (4 items, e.g., ''The conduct of the Board and (senior) management sets a good example in terms of ethical behavior''); feasibility (6 items, e.g., ''I have insufficient time at my disposal to carry out my tasks responsibly''); supportability (6 items, e.g., ''In my immediate working environment, everyone has the best interest of the organization at heart''); transparency (7 items, e.g., ''If a colleague does something which is not permitted, my manager will find out about it''); discussability (10 items, e.g., ''In my immediate working environment, there is adequate scope to discuss unethical conduct''); and sanctionability (10 items, e.g., ''If I reported unethical conduct to management, I believe those involved would be disciplined fairly regardless of their position''). The response scale ranged from 1 (strongly disagree) to 6 (strongly agree), and negatively worded feasibility item scores were reversed so that the higher the given score was, the stronger the virtue was. The copy of the scale can be found in English from the original paper of Kaptein (2008) and both in English and in Finnish from the dissertation of Huhtala et al. (2013). The means, standard deviations, and Cronbach's alphas of these virtues in every subsample are reported in Table 1.
The background variables to be controlled for in the analyses included gender (male/female), age (in years), and managerial level (upper management/management/upper middle management/lower middle management). Previous research has shown that younger employees are more likely to change their jobs (e.g., Mayer and Schoorman 1998;Moynihan and Landuyt 2008). Studies on the associations between gender and organizational position on the one hand and turnover on the other have yielded inconsistent results (see, e.g., the review of Moynihan and Landuyt 2008), with some studies showing that a higher organizational position relates to turnover, and some studies showing the opposite. Regarding gender, some studies have suggested that women are more likely to change jobs (e.g., Blau and Kahn 1981). However, a later labor economic finding has indicated that educated women and men resemble each other in turnover rates and patterns (see the review of Griffeth et al. 2000). Korabik (1992, 1995) found that situational factors affect turnover rather than gender or other person-centered explanations, and other factors, such as work-family responsibilities and marital status, explain the gender differences in turnover (Lee 2012). Some studies including meta-analyses have found that gender is a weak predictor of turnover (Bauer et al. 2007;Griffeth et al. 2000), but this viewpoint has been challenged by some researchers (e.g., Lee 2012; Sousa-Poza and Sousa-Poza 2007), as women's status in the labor market is continuously improving.

Analyses
Associations between background variables and managerial turnover (in Subsamples 1, 2, and 3) and background variables and the categories of reasons for turnover (Subsample 4) were investigated with v 2 tests (gender and managerial level), t tests and ANOVAs (age). After preliminary analyses, logistic regression analyses (background variables controlled for) were used to study whether the  eight virtues of ethical organizational culture predict managerial turnover. In Subsample 1, we studied whether ethical culture evaluations at T1 were associated with a manager's first change of job, between T1 and T2. In Subsample 2, the ethical culture virtues at T2 were set to predict a manager's first change of job between T2 and T3. In Subsample 3, ethical organizational culture at T1 was set to predict a manager's first change of job between T1 and T3. In Subsample 4, ANCOVA analyses (background variables controlled for) were used to study the associations between the perceptions of the virtues of ethical organizational culture (T1 or T2) and the first reason managers gave for their first job move between T1 and T3. The written reasons for changing jobs were thematically categorized into five content categories by two coders, who went through the data independently. Cohen's kappa coefficient (j = .94) showed excellent agreement of the categories between the two coders (Landis and Koch 1977). These five categories of reasons for a job change (presented later) were compared in further analyses to the group of those who stayed where they were. All of the analyses were performed with IBM SPSS statistics, version 22.

Descriptive Results
The majority of the respondents in this study did not change their job during the study period. In Subsample 1, 84 managers (19 %) changed their job and 367 (81 %) stayed in the same organization between T1 and T2. In Subsample 2, 45 (15 %) changed jobs between T2 and T3 and 263 (85 %) did not. Subsample 3 consisted of 125 (30 %) job changers and 285 (70 %) stayers between T1 and T3. Subsample 4 included 124 (32 %) managers who reported a job change and 258 (68 %) who did not leave their organization between T1 and T3. The results showed that gender and managerial level did not associate with job change in any of the four subsamples. Table 2 shows that those respondents who had changed their jobs were on average two years younger than the ones who had stayed in their organization in Subsamples 1 and 4, and three years younger in Subsample 3. In Subsample 2, job changers and stayers did not differ from each other in age. As a result of these findings, only age was included as a control variable in subsequent analyses.

Ethical Culture as a Predictor of Turnover
The results of logistic regression analyses in Subsamples 1, 2, and 3 are presented in Table 3. The results indicated, partly supporting our Hypothesis 1, that in all three subsamples, the virtues of congruency of supervisors and congruency of senior management were associated with manager turnover during the following two (Subsamples 1 and 2) or four years (Subsample 3). In addition, the virtues of discussability and sanctionability were found to predict manager turnover in Subsample 2.

Turnover Reasons
Of the managers who took part in Subsample 4 (n = 382), 124 had changed or left their job between T1 and T3 and gave a reason for their move. Five content categories were found on the basis of thematic categorization and they are listed in descending order of size in Table 4: (1) Lay-off (36 %), (2) Career/challenges (31 %), (3) Dissatisfaction with the job or organization (14 %), (4) Organizational change (10 %), and (5) Decreased well-being/motivation (9 %). These content categories were compared to the group of stayers in the analyses.
v 2 tests and ANOVAs were conducted to examine the relationships between the background variables and the job changers' and stayers' groups (Subsample 4). The results of the v 2 tests revealed no differences between the stayers and those who changed jobs in terms of gender or managerial level. Significant differences were found regarding age: managers in the groups of lay-off and stayers were on average older than the managers who were dissatisfied with the job or the organization, who explained their job change by decreased well-being or motivation, or who were seeking career opportunities or challenges (see Table 5). In subsequent analyses, age was used as a covariate.

Ethical Organizational Culture and the Reasons for Job Change
The results of ANCOVA analyses (see Table 6) showed that in general, perceptions of all the virtues of ethical organizational culture were statistically significantly associated with the categories of reasons given for changing jobs (p \ .001 and p \ .01). A closer look at the pairwise comparisons showed that, in line with quantitative results, stayers perceived many of the virtues of their organization's ethical culture more positively than the turnover groups with the exception of the group organizational change. More specifically, managers in the group of stayers gave higher evaluations for clarity, congruency of supervisors, and congruency of senior management than managers in the lay-off, career challenges, and dissatisfaction with the job/organization groups. They also perceived feasibility, supportability, and discussability more positively than the dissatisfaction with the job/organization and decreased well-being/motivation groups, and sanctionability more positively compared to the groups of career challenges, dissatisfaction with the job/organization, and decreased well-being/motivation.
The results also supported Hypothesis 2, showing that weaker ethical organizational culture was associated with turnover reasons related to factors pushing managers away from the organization (e.g., lay-off, dissatisfaction with the job or organization, decreased well-being/motivation). Those managers with decreased well-being/motivation or those who were dissatisfied with their job or organization gave the lowest mean scores on all the CEV sub-dimensions. Managers in the group dissatisfaction with their job or organization differed from the group of organizational change with more negative perceptions of clarity. The virtue of discussability was perceived more negatively in the groups of dissatisfaction with their job or organization and decreased well-being/motivation than in the group of stayers, and in the turnover groups of lay-off, career challenges, and organizational change. Group differences explained 14 % of the remaining variance of discussability, after age was controlled for. In other words, the turnover groups differed from each other most of all in their evaluations of discussability: the mean score differences between these groups were rather large.

Discussion
The present study aimed to fill the gaps in research regarding the associations between ethical organizational culture and turnover among managers in a 4-year longitudinal study. In addition, the reasons for manager turnover were investigated in relation to the virtues of ethical organizational culture. The advantage of using two different methods is that they can offer complementary information about the particular issue. The results revealed that the virtues of ethical organizational culture were linked to manager turnover and the reasons leading to it.

An Ethical Organizational Culture Keeps Managers in the Organization
In more detail, our first hypothesis was partly supported with the finding that perceptions of poor congruency of supervisors and congruency of senior management predicted turnover among managers in a two-year and fouryear period, and poor discussability and sanctionability predicted manager turnover in a 2-year period, after controlling for age. Our findings regarding congruency are in line with other studies that have shown that the ethical example set by top managers influences managers' job  Table 3 Logistic regression analyses of ethical organizational culture and managerial turnover (controlled for age) (T1-T2, n = 451; T2-T3, n = 308; T1-T3, n = 410) Corporate ethical virtues Subsample 1 (T1-T2) n = 84 (job changers) n = 367 (stayers) Subsample 2 (T2-T3) n = 44 (job changers) n = 264 (stayers) Subsample 3 (T1-T3) n = 125 (job changers) n = 285 (stayers) satisfaction, commitment, and turnover intentions (Kim and Brymer 2011), and that ethics-related action on the part of managers and supervisors-such as talking about ethics at work, keeping promises and commitments, informing employees, and modeling ethical behavior-produces organizational satisfaction (National Business Ethics Survey 2003). In addition, congruency of supervisors and congruency of senior management have been found to predict career goals related to job change among managers (Huhtala et al. 2013). Discussability, in turn, reflects the  ''An ever-increasing workload and due to that exhaustion'' ''I did not have motivation anymore and I was feeling exhausted'' possibilities of criticizing or debating ethical issues; it has been found to relate to career-ending goals among managers (Huhtala et al. 2013). If there is no room for ethical discussions, this moral muteness may lead to higher moral stress (Bird and Waters 1989) and possibly lead to a perceived breach in the psychological contract (O'Donohue and Nelson 2009). Poor sanctionability reflects a tolerance of unethical conduct and an undermining of the effectiveness of norms, as well as a lack of recognition of ethical conduct. In sum, for the managers in this study, the decision to change their job was affected not only by the perception that congruency was low in their organization, but also by the perception that their organization did not support discussion about ethical issues, and gave neither rewards for ethical conduct nor sanctions for unethical conduct.
With regard to the four abovementioned virtues, it can be assumed that when the organizational culture is less ethical, especially ethically oriented managers may experience difficulties in acting according to their own values, as P-O fit theory (Chatman 1989) suggests. In fact, it has been suggested that managers who consider ethical values to be important strive to work in a matching organization (Brown and Treviño 2006). A value misfit between the manager and the organization may lead to turnover or resignation, for instance as a result of an ethical conflict (Schwepker 1999;Thorne 2010). In addition, a less ethical organizational culture can be perceived as a breach in the psychological contract in itself, which can lead to resignation (O'Donohue and Nelson 2009). Managers can feel that unfulfilled commitments promised in the recruiting phase violate the psychological contract between them and their organization, and this can lead to turnover intentions (Collins 2010). Thus, by improving the ethical example set by higher managers, discussion about ethical issues, and rewards and sanctions for (un)ethical actions, it is possible to create a more favorable working environment for managers to operate in and to keep them in the organization.
The two virtues that had a steady impact on predicting managers' actual turnover throughout the four-year study period were congruency of supervisors and congruency of senior management, whereas poor discussability and sanctionability predicted manager turnover only in the later 2-year study period, between 2011 and 2013. The reasons why this connection emerged only in the later period of the study may be many. First of all, it may be that congruency is a steady predictor of turnover, whereas discussability and sanctionability may reflect the current economic and societal conditions more. The effects of the economic downturn that started in 2008 on the perceptions of ethical culture may have become visible in the 2011 questionnaire, underlining the importance of opportunities Table 6 Results of ANCOVA analyses for ethical culture in the categories of reasons for employee turnover (age as a control variable, N  to discuss ethical issues, and of fostering principles for the consequences of (un)ethical conduct. It is possible that the ethical organizational culture had actually become weaker between 2009 and 2011: as Huhtala et al. (2015) showed in their study on ethical organizational culture's influence on managers' well-being, one group of managers reported a steady decrease in ethical culture in a two-year time period. In addition, one partial explanation for the result in this study may be the smaller sample size in 2011 compared to 2009, which may have affected the selectivity of the respondents and consequently the results, and should be taken into consideration when interpreting the results.

Unethical Culture Pushes Managers to Change Their Job
The results showed that all the virtues of ethical organizational culture were associated with the different categories of reasons for turnover. Managers in the group of stayers perceived the virtues of their organization's ethical culture more positively than the turnover groups. The results supported the quantitative findings, showing that those managers who did not change their organization were more content with their ethical organizational culture than the turnover groups. In sum, managers who work in virtuous organizations are willing to stay there. Although the main interest of this study was on pushmotivated job change, the pull-motivated job change was also recognizable from the managers' answers. This finding was expected, since managers are likely to change jobs more often in order to advance in their career. Especially, those managers who have some special knowledge and skills may be headhunted by other companies, and social media have made networking and job search easier. Moreover, turnover may also have a weak but stable component (Semmer et al. 1996(Semmer et al. , 2014, implying that there is a tendency to more easily either stay or leave throughout one's career. As expected, the previous perceptions of the organization's ethical culture were not linked with pullmotivated quitting, as the pull-motivated managers had perceived their previous organization's ethical culture fairly positively. The groups of career challenges and organizational change included those managers who were attracted or ''pulled'' from their current positions for professional, financial, or personal advancement. Comparison of the mean scores showed that the turnover groups of career challenges and organizational change perceived many of the virtues more positively than the other turnover groups, although they were somewhat more critical in their perceptions than the group of stayers. This result was in line with the expectations that pull-motivated job changers may not be dissatisfied with their current job, as the pull motivation may represent ''the desire to try out something new; to expand one's skills, knowledge, and abilities; to seek a new challenge or the like are further possibilities. These do not require dissatisfaction with one's current job or organization.'' (Semmer et al. 2014, p. 75).
The results indicated that the group of career challenges did not differ from other turnover groups in their evaluations of the virtues of ethical culture, but they were more critical in their evaluations compared to the group of managers who did not change their organization in the four-year study period. Some managers in this group implied that they had already worked a long time in one organization and felt that it no longer provided them opportunities: ''I terminated a long career for one employer and moved towards new challenges and more responsible role.'' On the other hand, it seems that managers in this group were also dissatisfied with some things in the organizational culture, and career development was not the only reason for changing their job, as their comments showed: ''I wanted a change, to learn new, there was a good opportunity. I was also not satisfied with certain things in the previous work community'' or ''Career development was stuck, I got a better salary, supervisor was not good.'' The group of organizational change, on the other hand, differed from the other turnover groups with its more positive perceptions of the virtue of clarity and discussability. Many of the managers in this group had changed their tasks or organization due to acquisitions and mergers, but some also reported starting to work as an entrepreneur as a result of an organizational change. In this context, organizational change did not act as a push motivator driving the manager away from the organization: it is possible that in many cases these managers could negotiate their role in the future organization and participate in the change process. In addition, large organizational changes are usually planned in advance, phased in, and followed up, which might also explain the high scores given to the virtue of clarity, which refers to comprehensive and concrete expectations and norms for ethical behavior. Moreover, compared to organizational changes, which are usually implemented in as controlled a way as possible, temporary lay-offs, career challenges, and general dissatisfaction may cause more uncertainty among managers, which may explain why the ethical culture was perceived as lower among these groups.
The findings of this study supported the second hypothesis (H 2 ) claiming that a weaker ethical organizational culture will increase the probability for changing jobs because of push motivation, such as value conflicts or mismatches between the manager and their organization. Push-motivated job change groups were considered to include the following groups: dissatisfaction with the job/ organization, decreased well-being, and partly the group of lay-off, which included managers who were either forced to change their jobs, or left their current organization after negotiations and severance pay packages. In some cases, they had come under a lot of pressure to resign and had had little alternative but to do so. In Finland, upper-level managers often fall outside of the Act on contracts of employment: they, therefore, do not have the protection of the labor law and may be laid off at short notice. In consequence, they always risk facing the push factors that come on the initiative of the organization: being laid off or asked to resign. In the following sections, the results regarding the other two job change groups, dissatisfaction with the job/organization and decreased well-being, will be inspected in more detail, since these two groups were the most critical of all the virtues of ethical culture in their organization.
The reasons managers gave for their job change in the group of dissatisfaction with the job/organization, such as ''the misfit between own approach and the senior management's values'' (see Table 4), indicated that a job change was often preceded by a sense of incongruence between the manager's values and those of the organization. This finding is in line with studies showing that poor person-organization fit predicts turnover (O'Reilly et al. 1991), and that a misfit between values may create ethical conflicts that eventually lead to turnover, as Thorne (2010) suggested. Managers in this dissatisfaction with the job/ organization group rated all the virtues, except for transparency, lower than the managers in the group of stayers. Especially, the perceptions of congruency of supervisors and senior management, discussability, and sanctionability were fairly low in this group, indicating that the dissatisfied managers lacked an ethical example and role modeling from their managers and supervisors, lacked opportunities to discuss ethical issues, and perceived that unethical actions were not punished and ethical actions not rewarded in their organization. Another way to explain this result is a possible violation in the psychological contract between the managers and their organizations, which has previously been shown to lead to turnover intention among employees (Suazo et al. 2005;Sturges et al. 2005;Carbery et al. 2003) and managers (Shahnawaz and Goswami 2011). For instance, managers gave reasons for a job change such as ''dissatisfaction with the company's top management's acts'' and ''Being bullied at work by own supervisor,'' and this can be interpreted as disappointment with promises made at the recruiting stage. Unfulfilled promises like these, breaches in the psychological contract, have in fact been shown to lead to turnover intentions among managers (Collins 2010).
Another group of managers, who differed from the stayers with more negative perceptions of almost all the virtues, was the group of decreased well-being/motivation. However, this group did not differ statistically from the stayers in their perceptions of congruency of supervisors and senior management, like the group dissatisfaction with the job/organization did, although the mean score for congruency of supervisors was fairly similar in both of these turnover groups. A few studies have shown that the virtues of congruency of supervisors and senior management are important in preventing employee sickness absences (Kangas et al. 2015) and in promoting managers' well-being (Huhtala et al. 2011. However, our findings indicated that the managers who had changed their job due to exhaustion, work overload, or motivational issues gave the lowest mean scores of all the groups to the virtues of feasibility, supportability, transparency, discussability, and sanctionability, but not to congruency. Perhaps, in preventing managers from having too much work overload and keeping them motivated the ethical example of their supervisors is not enough; it is also important to provide adequate resources to enable ethical behavior, as well as encourage ethical conduct by rewards and sanctions for (un)ethical actions, and to provide possibilities of discussing ethical issues openly. In addition, the organization should encourage support and mutual trust among co-workers, which in managerial work could mean building up a supportive network with other managers.
The results of the present study were somewhat different for Hypotheses 1 and 2. The quantitative results on manager turnover in a longitudinal setting showed that four out of the eight virtues were related to turnover, whereas regarding the turnover groups, the results implied that all the virtues were perceived significantly differently between those managers who did not change their job and those managers who belonged to one or other of the turnover groups. One explanation for this may lie in the different ways of analyzing the data, as those managers who had changed jobs were divided into smaller subgroups based on the reasons they had given for their job change. Also the sample size was bigger in the first three subsamples, while the fourth subsample included only managers who did not change their job at all or managers who at the last measurement point provided some information in the openended questions about their reasons for changing their job. Those managers who took part in the whole survey may have been more motivated in the topic of the survey and in answering thoroughly, but they may have also been the ones who had something to criticize; especially, those managers who answered open-ended questions about their reasons for changing jobs may have had more negative experiences related to their move. However, the information was sought using different methods that were designed to complement and confirm each other, providing valuable information on the topic from a range of different points of view.

Limitations and Future Recommendations
This study has its limitations. First, the participants in this study were Finnish managers, and this sets some limits to the generalizability of the results. Second, the response rate in the first questionnaire round was fairly low. However, it has been suggested that for managerial data, a response rate of anything over 35 % is acceptable (Baruch and Holtom 2008). Third, other control variables, such as ethical climate and organizational culture in general, were not included in the survey. However, their inclusion would have made it very time-consuming and could therefore have yielded a lower response rate. In addition, those respondents who participated in the study through the four years may have been managers who were particularly interested in ethical topics or, conversely, had something to criticize. It is also important to note that although it is expected that ethical managers value ethical values in their organization, we did not include manager's self-evaluations of their own ethical leadership.
Future studies in ethical organizational culture and turnover could make use of additional objective measures. For instance, a company's financial ratios and performance could be studied as an antecedent of turnover at the same time as the ethical organizational culture, to find out what kind of additional effect an ethical organizational culture may have on turnover when these objective measures are included in the same analysis. In addition, it would be useful to find out whether a company's financial ratios and performance are associated with perceptions of how ethical an organization's culture is. Another fruitful direction would be to examine the (mis)match between managers' own personal ethical values and their organization's ethical values, to see if this fit or misfit plays a role in explaining subsequent turnover. Such a study would throw light on our previous assumptions that the person-organization value fit and psychological contract may explain the associations between ethical culture and turnover.

Conclusions
This study contributes to the literature by establishing the role of ethical culture in preventing manager turnover. Longitudinal setting with multiple measurement points was used, which has been a shortage in the ethical culture research. The results of the present study also provide new information on the reasons leading to quitting, which are still relatively unknown, and especially in relation to the values of an organization, which represent the ethical culture. For organizations, it is important to understand managers' motives for changing jobs and find out how to reduce the likelihood of them leaving, since managers' departure can be detrimental for organizations in various ways. Four out of eight virtues of ethical culture were related to fewer job changes in both the two-and fouryear follow-up periods, and all of the virtues were associated with the reasons for turnover. The ethical role modeling of supervisors and managers were most strongly associated with fewer turnovers among managers. Discussability and sanctionability were also associated with fewer job changes. From the examination of the reasons managers gave for turnover, one could see that the turnover group is not homogeneous, but there are several different reasons for leaving. Especially, those managers, who had perceived their organization's ethical culture as weakest, named job change reasons related to dissatisfaction with their job or the organization, or well-being or motivational issues. It seems that in preventing managers to leave, it is important that managers perceive that their own managers and supervisors ''walk the talk,'' that they have opportunities to discuss ethical issues, and that the sanctions and rewards for (un)ethical conduct are implemented justly and fairly. Therefore, it is important to nurture the virtues that reflect the ethical organizational culture to encourage all managers and supervisors to want to remain in their organization.