Managing Flexible Work Arrangements in Government: Testing the Effects of Institutional and Managerial Support

Previous Research on Telework

Telework refers to “a work arrangement that allows an employee to perform work, during any part of regular, paid hours, at an approved alternative worksite” (U.S. Office of Personnel Management [OPM], 2016). Quite a few studies have highlighted the potential benefits from telework (e.g., Stavrou, 2005; Stavrou & Kilaniotis, 2010). Telework might not only help workers balance their work and family life, but also reduce office costs, save costs associated with employee turnover, and attract and retain talented employees for organizations (Saltzstein, Ting, & Saltzstein, 2001; Stavrou, 2005). Others, however, have noted that telework can also pose various challenges to individuals and organizations (e.g., Baruch, 2001; Gajendran & Harrison, 2007; Greer & Payne, 2014). Teleworkers may experience psychological problems such as social isolation, insecure feelings about their job, and perceived deprivation of job opportunities due to their employment status (e.g., Baruch, 2001; Huws, 1993; Peiperl & Baruch, 1997). Scholars warned that such negative consequences should also be recognized and treated properly.

The more recent research has discussed effective strategies to implement telework successfully (Allen, Golden, & Shockley, 2015). Some studies indicated that support from the organization and supervisors plays a key role in the acceptance and administration of telework arrangements (e.g., Schmidt & Rosenberg, 2014; Shockley, Thompson, & Andreassi, 2013). Effective management of teleworkers (e.g., proper monitoring) and technical support that simulates face-to-face interactions are also important to telework effectiveness (Rein, 2014). In addition, others suggested that the moderate extent of telework, which balances face-to-face and virtual contact, will be most beneficial given that face-to-face interactions for knowledge sharing and social teleworking are as important to individuals as workplace flexibility and autonomy (Allen et al., 2015; Coenen & Kok, 2014; Gajendran, Harrison, & Delaney-Klinger, 2014).

Research on telework in public administration is relatively new. The common thread of the studies is that the family-friendly policies and programs that are designed to aid employees in balancing their work and family life may have positive associations with desirable work attitudes of employees (e.g., job satisfaction, organizational commitment; Caillier, 2012; Ezra & Deckman, 1996), but the relationship might vary across groups and programs (e.g., Kim & Wiggins, 2011; Saltzstein et al., 2001).

Telework in Public Management

There seem to be more concerns than promises about adopting the telework program in government. Despite presidential directives and legislative mandates over the past couple of decades, governmental organizations have made less progress on telework implementation than private industries (Green & Roberts, 2010). Scholars have been suspicious of the appropriateness of telework in public organizations, arguing that telework is more suitable for highly professional and flatter structures promoting autonomy and flexibility than for traditional bureaucratic organizations (Clear & Dickson, 2005; Cooper & Kurland, 2002). Indeed, U.S. OPM’s (2015) congressional report identified numerous barriers to telework in governmental agencies, for example, management resistance, technological challenge, and the nature of the work or position. Resistance by management is the most frequently reported obstacle to telework (U.S. OPM, 2015). Managers, from top to middle-level ranks, tend not to support telework or to have lack of understanding about how to organize, supervise, and control remote activities (Green & Roberts, 2010). In addition, lack of funding for constructing IT infrastructure, and the nature of jobs or positions that require face-to-face communication with clients or high levels of security and confidentiality were reported as the primary barriers to telework in government (U.S. OPM, 2015).

Nevertheless, Taskin and Edwards (2007), in their study of the Belgian public sector, have shown the possibility of successful adaptation of telework in public organizations. They argued that the sector might not be the critical factor that affects acceptance of telework if the following three contingencies hold. First, as in previous research, telework is more suitable for professional workers with higher discretion, performing complex tasks in flatter structures. Second, telework needs strategic support from management and transparent communication about the program to be adopted successfully. Finally, governmental agencies should significantly redesign their organizational rules so that the existing bureaucratic control rules, relying on the physical presence and visibility of workers, do not conflict with the new practices of control for telework arrangements (e.g., self-control, social control, proper use of electronic-based tools). More importantly, a trust relationship between teleworker and supervisor should be developed. If these conditions are met, telework in public organizations will enhance the advantages of traditional bureaucracy such as predictability, accountability, and control over workers (Taskin & Edwards, 2007).

The focus of this study centers on exploring the critical role of support from management in successfully implementing telework in public bureaucratic organizations. It tests the roles of managerial and institutional support in the relationship between telework and turnover intention of public employees. Previous research detected only a small gap between turnover intention of federal employees who telework and those who do not (Caillier, 2013). The finding may raise a question about the contribution of the program to the reduction of high costs associated with employee turnover. Some may argue that telework explains only a small portion of variance in employee turnover among numerous factors (Cotton & Tuttle, 1986). This research, however, was focused on the possibility that such seemingly minimal effect of telework may be because its costs offset part of its benefits. For instance, while telework might relieve its users of the double burden of family and work responsibilities, it also causes the negative psychological consequences to them, eventually reducing the positive effects of telework on their welfare. In this study, as Taskin and Edwards (2007) suggested, it was assumed that managerial and institutional support would preserve the benefits of telework by helping teleworkers avoid such negative feelings. This study adds to the literature by its empirical investigation of whether organizational and managerial support for teleworkers make a significant difference in the link between telework and turnover intention.

Telework and Turnover

Reports from previous research have indicated that telework can reduce turnover of employees (e.g., Konrad & Mangel, 2000; Perry-Smith & Blum, 2000; Stavrou, 2005). Employees benefit from greater flexibility and autonomy in managing their time and place to work and balancing their work and family responsibilities (Ezra & Deckman, 1996), which in turn contribute to lowering turnover of employees.

Social exchange theory is commonly adopted to theorize the relationship between telework and its individual-level outcomes. Social exchange theory explains the employee–employer relationship relying on the norm of reciprocity and social contracts (Rousseau, 1995). If employees perceive that their organization offers them preferable conditions or benefits, employees will reciprocate demonstrating higher loyalty and commitment to their organization (Guimaraes & Dallow, 1999). Teleworkers who were offered favorable benefits from their organization were likely to demonstrate higher satisfaction with their organization and lower intention to leave it. Conversely, employees who demand a flexible arrangement, but are not offered it, were likely to show higher intention to find an alternative job that provides the arrangement.

Much research has found evidence that supports social exchange theory. Telework was positively related to the reduction of turnover intention or actual employee turnover (e.g., Baltes, Briggs, Huff, Wright, & Neuman, 1999; Konrad & Mangel, 2000; Perry-Smith & Blum, 2000). Gajendran and Harrison (2007, p. 1528), from their meta-analysis of 46 studies, found that telework, which is perceived as “a positive signal” of trust and support for employees, increases the psychological attachment of employees to their organization, in turn lowering intention to leave (Rhoades & Eisenberger, 2002; Scandura & Lankau, 1997). In a similar vein, Caillier (2013) also found that although the difference was marginal, teleworkers tended to have lower intention to leave their organization than did nonteleworkers. McNall, Nicklin, and Masuda (2010) reported that the availability of flexible work arrangements was positively associated with lower turnover intention, mediated by greater enrichment from work to home. Kelliher and Anderson (2010) found evidence that employees who were allowed to telework showed a greater sense of commitment and obligation to their work and organization (Kelliher & Anderson, 2010). Grounded upon such theoretical perspective and empirical evidence, the following hypotheses were developed.

The Contextual Impacts: Organizational and Managerial Support

As Taskin and Edwards (2007) indicated, strategic support from management can contribute to the successful implementation of telework. In turn, the benefits from telework can influence the relationship between telework and turnover critically.

From an ecological perspective, some scholars (e.g., Bronfenbrenner & Ceci, 1994) argued that workplace flexibility is an attribute of an environment that promotes person–environment interactions that ultimately yield positive outcomes for employees, their families, and their organizations (Hill et al., 2008). Ecological systems theory explains that a flexible work arrangement is “a social and contextual attribute of workplaces constructed from both structural (the availability of policies and the basic nature of tasks performed, such as manufacturing vs. service) and interactional factors (supportive culture and leader-subordinate trust)” (Hill et al., 2008, p. 154). This suggests that contextual factors may play crucial roles in the relationship between flexible work arrangements and outcomes. Some research has shown that contextual factors including organizational size, industry, organizational sector, and union involvement significantly influence the relationship between flexible work arrangements and turnover (Stavrou & Kilaniotis, 2010). For instance, larger organizations, in general, are likely to have more resources for formal flexible work arrangements than smaller organizations; public organizations will be more committed to flexible work arrangements than private companies. Flexible work arrangements are more commonly found in the service industry than in the manufacturing industry, and unions are more supportive of flexible work arrangements for employee welfare (Bardoel, Tharenou, & Moss, 1998; Brewster, Mayrhofer, & Morley, 2004; Hogarth, Hasluck, Pierre, Winterbotham, & Vivian, 2000; Ingram & Simons, 1995; Morgan & Milliken, 1992).

This study focused on the potential impacts of managerial and institutional support on the relationship between telework and its outcomes. Organizational culture that encourages employees to utilize telework will prevent them from experiencing negative psychological stresses such as a feeling of isolation from their job and workplace, peer pressure, and lack of support from supervisors. In addition, managerial support from supervisors will help employees to take full advantage of telework. Employees may be afraid of negative repercussions, for example, being retaliated against for utilizing telework in terms of career advancement (e.g., lower visibility and less promotion opportunities), unless they perceive that their supervisors are supportive of such flexible work arrangements. Some scholars suggested developing a new model of leadership appropriate in teleworking environments (e.g., Dahlstrom, 2013; Taylor & Kavanaugh, 2005). Probably, the biggest challenge to teleworkers is being isolated from their peer groups and losing professional and social interactions with their colleagues (Taylor & Kavanaugh, 2005). They might also worry about their lower visibility to the manager, which could negatively affect their career advancement. Thus, the leadership style that develops support, communication, and trust in supervisor–subordinate relationships will be necessary for successful implementation of telework (Dahlstrom, 2013). Furthermore, such relationship-oriented leadership is needed more in government, given that the standardized and formalized processes typical in public bureaucracy could replace much task-oriented leadership behavior (Dahlstrom, 2013).

Saltzstein et al. (2001) emphasized that organizational understanding of employee-family demands was more likely to affect job satisfaction than was the availability of the policies themselves. Similarly, Ezra and Deckman (1996) also demonstrated that organizational understanding of family duties was positively related to employee satisfaction with work–family balance. Kim and Wiggins (2011) called for organizational culture that is supportive of work–family balance to better utilize work–family balance polices and take fuller advantage of the policies. Taskin and Edwards (2007) argued that managerial supports and cultural changes within organizations are especially important in public bureaucratic settings because these tend to have less favorable conditions for adopting telework, characterized by strict rules, hierarchical structures and decision making, and control and surveillance. Nonetheless, when individuals internalize collective norms and values, and build a trust relationship with management, telework arrangements will work well even in bureaucratic settings, reinforcing the “bureaucratic virtues” such as predictability, accountability, and control over workers (Taskin & Edwards, 2007, p. 204). Based on these arguments, employee perception of institutional and managerial support is assumed to decrease their intention to leave the organization. H3 and H4 examine the two-way interaction effects between organizational contexts and telework, while H5 tests the three-way interaction effect.

Various factors (e.g., social, organizational, job, and personal characteristics) are likely to affect the turnover intention of an individual (Cotton & Tuttle, 1986; Mobley, 1982) and these factors need to be controlled. In much research on turnover intention and turnover, minority status and gender have been controlled (Moynihan & Landuyt, 2008; Whitford & Lee, 2015). However, the findings regarding those relationships appear to be inconsistent. Some studies found minority employees more likely to leave their current organization than Whites due to some reasons (e.g., discrimination), but others revealed a negative (or no significant) relationship between minority status and voluntary turnover (Lee & Whitford, 2008; Shields & Price, 2002). The relationships between gender and turnover and turnover intention are also unclear. Whereas some research (e.g., Cotton & Tuttle, 1986; Kellough & Osuna, 1995) found that women are more likely to leave their organization than men, others found insignificant or even negative relationship between women and turnover (Stumpf & Dawley, 1981). The more recent study by Moynihan and Landuyt (2008) contended that the higher probability of female turnover may not be true any longer given the changing patterns of female employment and their preference for public employment. In regard to age and tenure, Cotton and Tuttle (1986), in their meta-analysis, found negative relationships between age and tenure, and turnover. Moynihan and Pandey (2008), however, argued that the relationship between age and turnover intention might vary depending on whether it is long-term (negative) or short-term (insignificant) turnover intention. Supervisory status and disability status might be related to access to alternative jobs (Cotton & Tuttle, 1986; Whitford & Lee, 2015). Individuals who hold a higher position in an organizational hierarchy are likely to have more chances to find alternative jobs due to their experience and ability. Conversely, people with disability may suffer from limited access to alternative jobs.

In addition, this study also controls for the perception of work involvement as a proxy measure of social isolation, as well as individual characteristics. If employees perceive a higher level of teamwork or cooperation in their work group, they are less likely to leave (Mobley, 1982). Figure 1 displays the hypothesized relationships.

OPEN IN VIEWER

Figure 1.

The hypothesized model.

Sample and Data Sources

A sample from the federal workforce was analyzed. The federal government, the largest single employer in the United States, announced strong support for creating a new working environment by signing the Telework Enhancement Act in 2010 and has reported the status of telework in the federal government on a regular base (U.S. OPM, 2013a). Since enactment of this law in 2010, federal workplaces have been significantly transformed, for instance, the number of federal employees who are eligible and who participate in the program continues to grow (U.S. OPM, 2013a).

The sample data were drawn from two sources: the U.S. OPM 2013 congressional report, and the Federal Employee Viewpoint Survey (FEVS), collected in the same year. Since the 2010 Telework Enhancement Act, U.S. OPM provides annual reports to U.S. Congress, for which information on the status of telework programs of executive branch agencies is collected and summarized (U.S. OPM, 2013a). The 2013 FEVS was electronically conducted among federal employees including full-time and part-time, permanent, nonseasonal employees (U.S. OPM, 2013b). A total of 376,577 out of 781,047 employees from 81 federal agencies completed the survey, resulting in a response rate of 48.2% (U.S. OPM, 2013b). The survey includes questions regarding turnover intention, job satisfaction, supervisory support, and telework status of respondents, which were extracted and merged into the data for analysis.

Dependent Variable

Independent Variables

Control Variables

All the control variables in the models were measured using the responses to the relevant survey questions. The supervisory status variable was measured based on the responses to the question, “What is your supervisory status?” Supervisors, managers, and executives were coded “1”; nonsupervisors and team leaders, as “0.” Employees with disability were coded “1” and others were coded “0.” The survey question only asks if individuals had a disability; the types of disability were not reported. Tenure and age are categorical variables. Tenure was measured on a 3-point scale, ranging from “1” to “3.” Here, “1” indicates that the respondent’s tenure with the federal government was 5 or fewer years, “2” indicated 6 to 14 years, and “3” indicated 15 or more years. In a similar way, the age variable was coded on a 4-point scale: “1” indicated <40, “2” indicated 40 to 49, “3” indicated 50 to 59, and “4” indicated 60 years or older. Other demographic variables included gender and minority status. A female employee was coded “1,” while a male employee was coded “0.” A racial/ethnic minority employee was coded “1,” while a White employee was coded “0.” A work involvement variable was developed combining two survey items, which inquired of cooperation and knowledge sharing of the respondent with coworkers. The responses to the two questions were averaged. The Cronbach’s alpha was .89. Table 1 provides descriptive statistics of the variables.

OPEN IN VIEWER

Table 1.

Descriptive Statistics.

Variable	M	SD	Minimum	Maximum	Unit of analysis
Turnover	0.31	0.46	0	1	Are you considering leaving your organization within the next year? Yes = 1 (31%); No = 0 (69%)
Age	2.32	1.00	1	4	What is your age group? 1 = “Under 40”; 2 = “40-49”; 3 = “50-59”; 4 = “60 or older”
Tenure	2.11	0.81	1	3	How long have you been with the federal government (excluding military service)? 1 = “5 or fewer years”; 2 = “6-14 years”; 3 = “15 or more years”
Supervisor	0.18	0.38	0	1	What is your supervisory status? 1 = “Nonsupervisor/team leader”; 0 = “Supervisor/manager/executive”
Disability	0.15	0.36	0	1	1 = People with disability (15%) 0 = No disability (85%)
Minority	0.33	0.47	0	1	1 = Minority (33%) 0 = Whites (67%)
Sex	0.40	0.49	0	1	1 = Women (40%) 0 = Men (60%)
Teleworkers	0.24	0.43	0	1	1 = Teleworkers (24%) 0 = Nonteleworkers (76%)
Nonteleworker by barrier	0.56	0.50	0	1	1 = Nonteleworker by barrier (56%) 0 = Teleworker and nonteleworker by choice (44%)
Teleworkers (% by agency)	13.48	13.86	0	96	Teleworkers / the number of employees by agency × 100 (%)
Managerial support	3.97	0.96	1	5	Responses to three questions were averaged (Please see the appendix to find specific questions)
Work involvement	3.78	0.92	1	5	Responses to two questions were averaged (Please see the appendix to find specific questions)

Note. The survey weights were adjusted for the analysis of the data sampled from the federal workforce.

Analytic Models and Methods

The research question was whether telework, along with managerial and institutional support for it, could reduce employee intention to leave. To better understand the characteristics of the sample, the data were grouped into three categories by telework status—teleworkers, nonteleworkers by barrier, and nonteleworkers by choice. Then, to compare group differences (including turnover intention), multivariate analysis of variance (MANOVA) tests were conducted. Table 2 displays the results. The three groups were significantly different in terms of turnover intention and demographic characteristics. Nonteleworkers due to barrier (M = 0.34) reported the highest turnover intention, while nonteleworkers by choice (M = 0.27) reported the lowest. Over half of the teleworkers (M = 0.53) were women and the proportion of women of the group who chose not to telework (M = 0.41) was lowest, suggesting that women are more likely to telework than men if they are allowed. The proportion of minority employees in the group of teleworkers (M = 0.31) was lower than that in other groups. Interestingly, the proportion of people with disability in the group of teleworkers (M = 0.09) was lower than that in the group of nonteleworkers by barrier (M = 0.16) and by choice (M = 0.11). This appears inconsistent with the common expectation that people with disability are more likely to telework. Given the nature of managerial duties, the highest proportion of supervisors (M = 0.23) among the three groups, was found in the group of those who chose not to telework. Older employees, and those with longer organizational tenure, were less likely to telework by choice, suggesting that they preferred traditional work settings.

OPEN IN VIEWER

Table 2.

MANOVA Testing for Differences in Means by Telework Status.

	Telework status
	Nontelework by barrier	Teleworker	Nonteleworker by choice	Total	F(df = 7)
	M (SD)	M (SD)	M (SD)	M (SD)
Turnover intention (yes = 1; no = 0)	0.34 (0.48)	0.29 (0.46)	0.27 (0.44)	0.31 (0.46)	441.09***
Women (proportion)	0.44 (0.50)	0.53 (0.50)	0.41 (0.50)	0.48 (0.50)	1,492.98***
Minority (proportion)	0.35 (0.48)	0.31 (0.46)	0.34 (0.47)	0.33 (0.47)	201.74***
People with disability (proportion)	0.16 (0.37)	0.09 (0.28)	0.11 (0.31)	0.12 (0.33)	1,255.57***
Supervisors (proportion)	0.19 (0.40)	0.20 (0.40)	0.23 (0.42)	0.20 (0.40)	139.60***
Age	2.40 (0.97)	2.36 (0.97)	2.63 (0.99)	2.41 (0.98)	891.87***
Tenure	2.14 (0.80)	2.24 (0.80)	2.32 (0.79)	2.25 (0.80)	685.28***

Note. Fs indicate univariate Fs from a multivariate analysis. The survey weights were adjusted for the analysis of the data sampled from the federal workforce.

*

p < .05. **p < .01. ***p < .001.

To test the hypotheses, the models were analyzed statistically. The agency fixed-effects logistic regressions were used to test the relationship between telework status and turnover intention. The turnover intention of a respondent was measured on a binary scale. To test the moderating effects of managerial and institutional support, multiplicative variables were generated. Multiplicative variables and their constituents are likely to be highly correlated, which may lead to multicollinearity. To address that concern, all independent variables and control variables were centered to their mean values before being entered in the models. The correlations between variables are reported in Table 3.

OPEN IN VIEWER

Table 3.

Bivariate Correlations.

	Turnover	Age	Tenure	Supervisor	Disability	Minority	Sex	TW	NTWB	TW%	Managerial support
Age	−0.03
Tenure	−0.04	0.49
Supervisor	−0.02	0.12	0.18
Disability	0.05	0.10	−0.06	−0.02
Minority	0.02	−0.05	−0.02	−0.03	0.03
Sex	−0.01	−0.01	0.08	−0.07	−0.10	0.11
TW	−0.03	0.01	0.05	0.00	−0.06	−0.03	0.13
NTWB	0.06	−0.07	−0.08	−0.01	0.07	0.01	−0.12	−0.72
TW%	−0.05	0.05	0.07	−0.02	−0.05	0.00	0.09	0.38	−0.33
Managerial support	−0.24	−0.02	−0.05	0.04	−0.05	−0.05	−0.01	0.12	−0.14	0.07
Work involvement	−0.21	0.04	0.00	0.07	−0.04	−0.05	0.07	0.08	−0.11	0.07	0.48

Note. The survey weights were adjusted for the analysis of the data sampled from the federal workforce. All the correlation coefficients are statistically significant. TW = teleworkers; NTWB = nonteleworkers barrier.

The sample weights of all estimates in this study were adjusted to reflect population characteristics properly. The sample was randomly selected, then stratified by supervisory status and organizational membership. The differing response rates among groups (e.g., different demographic groups and supervisory groups) may result in unequal representation of responses across groups. For example, supervisor responses were overrepresented and agency effects may affect the results due to a natural clustering effect by agency. ⁴ To avoid such errors and to ensure external validity of the findings, the data were weighted and analyzed.

The models include two different levels of variables. The main independent variable, the dependent variable, and demographic variables are individual-level variables, while institutional support, which is one of the moderating variables, is an agency-level variable. Because federal employees are clustered into agencies or programs that share policies and procedures, the agency effects were controlled by developing fixed-effect models. In general, multilevel models are used to test relationships among variables that apply at two or more levels (Rousseau, 1985, p. 16). However, multilevel modeling may be problematic in cases that the models include a small number of upper level units (number of agencies in this study; Mohring, 2012). In such cases, a fixed-effects model provides an effective alternative in analyzing the nested data (Huang, 2016; Mohring, 2012). Quite a few scholars have indicated that the fixed effects approach is a robust modeling strategy; therefore, it has often been used in social science research to explain clustering effects (Chaplin, 2003; Clarke, Crawford, Steele, & Vignoles, 2010; Huang, 2016; Murnane & Willett, 2011).