Expatriate work role engagement and the work–family interface: A conditional crossover and spillover perspective

COR theory

Central to understanding the crossover and spillover processes through which the family sphere influences expatriate work outcomes are the various resources that expatriates have at their disposal (Lazarova et al., 2010). Thus, we invoke COR theory (Halbesleben et al., 2014; Hobfoll, 2001) to better articulate these processes. COR theory is a motivational theory that explains human behavior based on the evolutionary need to protect current and obtain new resources (Hobfoll et al., 2018). The theory holds that individuals must invest resources to cope with the presence of stressors, which threaten existing resources, and that unsuccessful coping may lead to a spiral of resource loss.

COR theory categorizes resources based on their origin and their degree of transience (Hobfoll, 2002; Ten Brummelhuis and Bakker, 2012). First, resources can be either located outside the self (i.e. contextual) or they can be proximate to the self like personal traits or energies (i.e. personal). Second, resources vary depending on whether they are transient and fleeting (i.e. volatile), or whether they are durable and stable (i.e. structural). Specifically, a partner’s family role adjustment serves as a contextual resource for the expatriate. While the partner’s family role adjustment may be stable in the short term, it can shift over time as the partner navigates the changes to the family domain while abroad, thus giving it a volatile nature.

COR theory also discusses the role of key personal resources (Halbesleben et al., 2014; Hobfoll, 2002). Unlike other resources, which may be finite and have more limited utility and scope (e.g. energy, skills, material and social support), key personal resources have far-reaching benefits and serve to unlock, enable and maximize the use of other resources (Halbesleben et al., 2014; Ten Brummelhuis and Bakker, 2012). They include certain personality traits (e.g. conscientiousness, emotional stability) and stable personal attributes (e.g. self-efficacy, emotional intelligence, optimism) that are versatile and durable enough to substitute other resources and facilitate the accumulation of new resources for attaining one’s goals (Halbesleben et al., 2014). Key resources provide an explanation for why some individuals are better able than others to cope with stressful circumstances and accumulate new resources (Ten Brummelhuis and Bakker, 2012). Owing to their versatility and durability, key personal resources should be particularly salient when individuals encounter major role transitions, like during expatriation.

General self-efficacy

A key personal resource that should protect and grow expatriates’ resource endowment abroad is self-efficacy. Here, we focus on general self-efficacy, which reflects a “belief in one’s capabilities to organize and execute the course of action required to produce given attainments” (Bandura, 1997: 3). While self-efficacy is often treated as a domain-specific attribute that changes with the situation (e.g. Gist and Mitchell, 1992), in line with our COR theory perspective and the definition of key personal resources we are particularly interested in self-efficacy as an attribute that generalizes across tasks and situations.

There are two reasons for focusing on general self-efficacy in the expatriate context. First, self-efficacy has long been recognized theoretically and empirically as an important input to expatriate success (e.g. Black et al., 1991; Shaffer et al., 2016; Wu and Bodigerel-Koehler, 2013). Black and colleagues’ (1991) model of expatriate adjustment, which has been the dominant theoretical foundation for an extensive body of research on expatriates, positions self-efficacy as an individual difference critical for adjustment—an assertion later supported in a meta-analysis by Bhaskar-Shrinivas et al. (2005). Self-efficacy beliefs play a particularly central role during expatriation, as expatriates need to perform across a range of new situations, including new work roles, changed family dynamics and a novel host-country context (e.g. Shaffer et al., 2016). Navigating the far-reaching changes to all life domains owing to relocation is easier when expatriates have the confidence to embrace the challenges and quickly bounce back from adversity. Indeed, self-efficacy captures differences in individuals’ tendency to view themselves as capable of meeting different task and relational demands in a broad array of novel circumstances, persevere in the face of stressful and demanding situations (Bandura, 1986; Chen et al., 2001), succeed in attaining their goals (e.g. Phillips and Gully, 1997) and proactively shape their environment (Niessen et al., 2016; Parker et al., 2010).

Second, despite the recognized importance of self-efficacy for expatriate success, we know little about how it interacts with influences from the expatriates’ family domain. We believe general self-efficacy is a critical boundary condition for crossover and spillover during expatriation. As a cognitive appraisal of one’s confidence to succeed, self-efficacy complements the primary focus on affective and behavioral mechanisms through which work and life domains are thought to shape each other (Leslie et al., 2019; Neff et al., 2013). Further, it helps expand the primarily affective lens that expatriate scholars have adopted to study expatriate experiences (Haslberger et al., 2013). In short, we conceive expatriate general self-efficacy as a key personal resource that conditions the availability and use of expatriates’ other resources and determines when and in which direction expatriates are more or less affected by crossover and spillover.

Crossover of affect: From partner family role adjustment to expatriate family role adjustment

Crossover is an interpersonal process through which negative and positive experiences and states experienced by one person are transmitted to another person in the same social environment (Bolger et al., 1989; Westman, 2001). To that end, research has provided evidence of crossover of burnout, anxiety, adjustment, satisfaction, engagement and happiness between partners (e.g. Carlson et al., 2019; Westman et al., 2009). Crossover has also been studied in the expatriation literature, with a focus on how the partner’s adjustment crosses over to the expatriate’s adjustment. This focus is not surprising considering that studies of antecedents to expatriate adjustment have dominated expatriation research (Takeuchi, 2010). Expatriate adjustment is most often defined as the “degree of a person’s psychological comfort with various aspects of a new setting” (Black and Gregersen, 1991: 498).

Following this conceptualization, studies have provided clear evidence for crossover of adjustment between the partner and the expatriate (e.g. Black and Stephens, 1989; Takeuchi et al., 2002). However, past studies have only accounted for adjustment to the host-country environment and do not examine adjustment to family life during expatriation. As a result, Lazarova et al. (2010) stressed the need to consider adjustment to the family role to understand the interplay between family members during expatriation. Similar to Black and Gregersen’s (1991) definition of adjustment, family role adjustment concerns the feeling of comfort, in this case toward the tasks and responsibilities of the family role and toward the interactions with family members (Shaffer et al., 2016). Accordingly, and based on COR theory, we conceive the partner’s family role adjustment as a focal contextual resource in the crossover process because it captures the partner’s experiences strictly within the family domain.

We propose that through a process of crossover the partner’s family role adjustment is transmitted to the expatriate. Positive affective states of other individuals serve as contextual resources (Hobfoll, 2002; Ten Brummelhuis and Bakker, 2012) and are thought to cross over directly from one person to the other owing to an empathic reaction, similar to mood contagion (Westman, 2001). Specifically, when one partner is well adjusted to the family role, that partner’s positive affective state should cross over to the other partner and facilitate his/her own family role adjustment (Lazarova et al., 2010). In other words, a partner’s family role adjustment is a contextual resource that should cross over to the expatriate, thereby facilitating the expatriate’s resource accumulation and ability to cope with stressors during the expatriate experience. While, to the best of our knowledge, there are no empirical studies on the crossover of family role adjustment, there is strong empirical support for the crossover of adjustment to the host-country environment (Bhaskar-Shrinivas et al., 2005; Takeuchi et al., 2002).

Further, we contend that expatriate self-efficacy acts as another resource that determines when expatriates are particularly affected by their partner’s family role adjustment. In line with the equifinality concept in COR theory (i.e. resources may be substituted for one another to reach the same goal; Halbesleben et al., 2014), individuals with high levels of personal resources like self-efficacy can replace (or substitute) contextual resources to still increase their resource endowments and achieve their goals. Key resources like self-efficacy enable individuals to mobilize other resources including energy or mental resilience (Ten Brummelhuis and Bakker, 2012). Therefore, expatriates with high levels of self-efficacy should have more personal resources at their disposal to meet the various demands posed by novel situations (Cast and Burke, 2002; Chen et al., 2001). As a result, they should be more capable of adjusting to their own family role, irrespective of whether or not their partners are adjusted to their family role.

When expatriates’ self-efficacy is low, we expect them to be more affected by the partner’s family role adjustment. The resource substitution concept of equifinality suggests that expatriates with low self-efficacy can benefit from their partners’ high levels of family role adjustment to adjust to their own family role and meet other family members’ needs. However, expatriates with low self-efficacy may also be more susceptible to low levels of partner family role adjustment. According to COR theory, a partner’s lack of role adjustment threatens expatriates with resource loss (Halbesleben et al., 2014; Hobfoll, 2001). For example, partners who struggle with adjusting to their new family role can drain expatriates’ energy and may require expatriates to spend more time and energy attending to their partners’ needs at the expense of other family members. Expatriates with low levels of self-efficacy should also be less able to regulate the risk of negative emotional contagion from their partners (Hatfield et al., 1993).

In sum, we expect that expatriate self-efficacy moderates the crossover between partner family role and expatriate family role adjustment. Specifically, for expatriates with high levels of self-efficacy there should be no crossover. However, without high levels of self-efficacy there will be a positive crossover between partner and expatriate family role adjustment. We propose:

The link between affect and motivation: From expatriate family role adjustment to expatriate family role engagement

We expect that the more expatriates are adjusted to their family role the more they are engaged in it. This is in line with COR theory’s notion that resources co-travel in caravans such that resources reinforce each other and facilitate the accumulation of other resources (Hobfoll et al., 2018). Specifically, when expatriates experience higher levels of family role adjustment, they will feel more comfortable in their role as a family member, which in turn enables them to exert more effort and become engaged in their family role (Lazarova et al., 2010). As Kahn (1990, 1992) stated, the conditions for engagement are psychological meaningfulness, psychological safety and psychological availability, and all three can be facilitated by adjustment. Family role adjustment thus provides expatriates with affective resources that reduce the uncertainties within the family context, enabling expatriates to be more engaged in the family. Empirically, Shaffer et al. (2016) provided some initial evidence for a positive link between expatriate family role adjustment and their family role engagement. Taken together, we propose:

Our previous theorizing implies that, within the boundaries of the family domain, there is an indirect relationship between partner family role adjustment and expatriate family role engagement via expatriate family role adjustment. This represents the process through which affect (i.e. the crossover of adjustment) translates into motivation (i.e. expatriate family role engagement). However, since we expect expatriate self-efficacy to moderate the crossover of adjustment, we propose that the indirect effect is also conditional on the level of self-efficacy (i.e. first-stage moderated mediation). Because expatriates with high levels of self-efficacy have more energy and cognitive resources at their disposal to meet situational demands (Cast and Burke, 2002; Chen et al., 2001), they should be more capable of attending to other family members’ needs, adjust to them and, in turn, become engaged in their family role irrespective of whether or not their partners are adjusted to their family role.

However, when expatriate self-efficacy is low, expatriates depend on contextual resources—including their partner’s family role adjustment—to adjust to their own family role, and in turn become engaged in their family role. Thus, when partners are well adjusted to their family role, expatriates with low self-efficacy may benefit from their partners’ positive affective state. At the same time, when partner family role adjustment is low, expatriates with low self-efficacy should be less able to avoid the depletion of their own resources (Hobfoll et al., 2018). In sum, at low levels of expatriate self-efficacy, partner family role adjustment will more likely relate to expatriates’ engagement in their family role via their own family role adjustment. Thus:

Spillover of motivation: From expatriate family role engagement to expatriate work role engagement

Spillover theory (Crouter, 1984) provides a theoretical rationale for how experiences in one life domain can influence experiences in another domain. It conceptualizes an intrapersonal process in which an individual’s attitudes and behaviors transfer from one role to another (Carlson et al., 2006). While the idea of spillover is widely accepted, the exact way in which the work and family domains interact has been a subject of debate, and the theoretical and empirical literature has suggested the full gamut of relationships, ranging from negative to non-existent to positive (e.g. Allen and Martin, 2017; Edwards and Rothbard, 2000; Rothbard, 2001).

Although the arguments vary depending on the variables studied, those advancing the idea of a negative interface between the work and family domains tend to focus on a resource-depleting effect of engaging in both work and family roles (Edwards and Rothbard, 2000; Greenhaus and Beutell, 1985; Takeuchi, 2010). Dominating early spillover research, this perspective suggested trade-offs between engaging in the two domains, and that engaging in one role comes at the expense of engaging in another (Byrne and Canato, 2017; Marks, 1977). Such tradeoffs imply the need to choose between domains and that, by extension, choosing to devote more to one domain inevitably leads to devoting less to the other (Edwards and Rothbard, 2000).

By contrast, those proposing a positive interface between the work and family domains (Edwards and Rothbard, 2000; Greenhaus and Powell, 2006) emphasize that simultaneously engaging in work and family may lead to “enrichment” rather than conflict. Engagement in one domain can generate pleasurable experiences and well-being, and it may create additional resources that can be applied to the other role (Edwards and Rothbard, 2000; Greenhaus and Powell, 2006). Indeed, whereas time is finite (this idea is one of the cornerstones of the conflict/depletion argument), engagement is not, and it should not be assumed that engaging in one role automatically means lower engagement in another role (Rothbard, 2001).

In sum, when it comes to engagement in work and family roles, research has advanced two contradictory propositions. On the one hand, individuals with family duties who are very engaged in their family role may deplete their physical, emotional and cognitive reserves, thus reducing their engagement at work. On the other, those heavily invested in their family role may generate and transfer positive energy from their family to their work, thereby increasing their engagement at work (Byrne and Canato, 2017). Returning to research on expatriates, Lazarova et al. (2010) specifically considered the relationship between expatriate family role engagement and work role engagement but, given the myriad possible relationships, their article included only a general spillover proposition “without specifying whether the relationship will be positive or negative” (2010: 106) and explicitly encouraged future research to examine individual variables that may serve as moderators of the relationship. This is consistent with the broader work–life balance literature (e.g. Greenhaus and Powell, 2006; Leslie et al., 2019), which has also called for examining possible moderators of the spillover between the work and family domains.

We expect that the key personal resource of general self-efficacy is a salient moderator because it should facilitate intrapersonal functioning across the family and work domains, both of which compete for resources. In other words, self-efficacy should determine whether family role engagement enables or frustrates work role engagement. Following COR theory (Halbesleben et al., 2014; Hobfoll, 2002), an individual’s key personal resources serve to unlock and enable other resources at work and at home. Specifically, expatriates with high levels of self-efficacy should be able to mobilize additional cognitive energy and positive mood (Ten Brummelhuis and Bakker, 2012). This should help them experience the family role in positive terms and, in turn, provide them with the necessary resources to engage in their work role. Indeed, new resources may be generated that can be applied to the work role (Greenhaus and Powell, 2006; Ruderman et al., 2002). Individuals are more likely to apply a resource generated in one role domain to another role domain when they believe that applying the resource will have positive consequences (Greenhaus and Powell, 2006). We would expect individuals to more likely perceive benefits of a particular resource for another domain (e.g. work) if they are confident that they can succeed in the focal role domain (i.e. family). All this should increase the likelihood that expatriate family role engagement positively spills over to the work domain.

By contrast, at low levels of self-efficacy the expatriates’ resources are more likely to be outstretched or exhausted owing to being engaged in the family role, thereby leading to depletion. This can lead to frustration and fatigue, which is unlikely to generate new energy that can be deployed to engage in the expatriate’s work role. An individual is also less likely to perceive benefits of transferring particular resources, such as interpersonal qualities and the ability to multitask (Ruderman et al., 2002) to another domain (e.g. work) if the individual does not feel confident to succeed in the focal role domain (i.e. family). In sum, we would expect positive spillover between family role and work role engagement at high levels of general self-efficacy, and negative spillover between family role and work role engagement at low levels of general self-efficacy. We propose:

Our previous arguments imply a second-stage moderated mediation where there is an indirect effect between expatriate family role adjustment and expatriate work role engagement through their family role engagement, conditional on expatriate general self-efficacy. This comprehensively captures the intrapersonal process of spillover between an expatriate’s family and work roles. That is, affective family role experiences (i.e. expatriate family role adjustment) translate into a spillover of motivation from their family role to the work role (i.e. spillover of expatriate family to work role engagement). In line with Hypothesis 4, we therefore expect this indirect relationship to be positive when general self-efficacy is high and negative when general self-efficacy is low. Hence:

Taken together, Hypotheses 4 and 5 suggest that whether expatriate family role engagement spills over to positively or negatively affect work role engagement—and whether expatriate family role adjustment positively or negatively relates to expatriate work role engagement via family role engagement—depends on the moderating effect of expatriate self-efficacy. This is an example of a disordinal interaction effect where the sign of the relationship between expatriate family role and work role engagement is contingent on the value of the moderator (see Cohen et al., 2003). Thus, we do not expect a significant relationship between expatriate family role and work role engagement without considering the moderator. As a result, we also do not expect a simple serial indirect effect from partner family role adjustment to expatriate work role engagement without accounting for the moderating role of self-efficacy.

Sample and procedure

Our study was part of a larger data collection conducted to assess the work and family experiences of global employees and their partners. Data for the present study came from seven organizations across two time points. Four of these were multinational companies, one was an expatriate association, one was a university alumni association and one was a foreign affairs ministry. We contacted executives in each organization and requested our first online survey to be forwarded to the global employees who were currently employed with the company/embassy, were members of an expatriate organization or were university alumni. At the beginning of the survey, participants received definitions for the terms ‘corporate and self-initiated expatriates’ and selected the one that corresponded to their current situation. The participants from the foreign affairs ministry were all diplomats at a foreign posting.

A total of 1477 expatriates participated in our study at Time 1. Since the organizations themselves distributed the survey, we could not track how many invitations were sent out, thus making it impossible to calculate a response rate. To test our model, we needed multi-source data from expatriates and their partners. Thus, the expatriates who were married or in a committed relationship were asked to provide their partners’ contact information. We then contacted the partners and asked them to complete a separate online survey. We informed expatriates and partners that survey participation and provision of contact details for their significant other is voluntary and that their data would be kept confidential. Respondents were entered in a lucky draw to win a US$100 gift card. Out of the 1477 expatriates, we obtained contact information for a total of 564 partners, who were each sent a partner survey at Time 1. Of these, we received 206 completed surveys (a response rate of 37%) with the information necessary to match the expatriate and partner data. To mitigate the threat of common method bias (Podsakoff et al., 2003) and assess the temporal ordering of the crossover and spillover processes, we not only collected data from multiple sources but also asked expatriates to complete a second follow-up survey approximately three months after completion of the data collection at Time 1. In short, while we used the expatriate and partner surveys at Time 1 to measure partner family role adjustment, expatriate family role adjustment and expatriate general self-efficacy, expatriate responses at Time 2 were used to measure expatriate family role engagement and work role engagement.

Of the 206 expatriates at Time 1, we received 105 completed Time 2 surveys for a retention rate of 51%. As a result, the final sample is comprised of 105 couples, which we used to test all our hypotheses. We checked for non-response bias across waves, by comparing expatriates with matching partner data who responded only at Time 1 (N = 101) with those who responded to both surveys across several demographic variables. The respondents who completed both waves did not differ significantly from those who only participated in the first wave. Thus, the sample at Time 2 is likely representative of the larger sample at Time 1.

Expatriates were mostly male (79%), had been in the host country for a little over two years on average and 58% had at least one child accompanying them abroad. Their mean age was 44 years and they had been married or in a committed relationship for an average of 16 years. The sample consisted of 40 diplomats, 58 corporate and seven self-initiated expatriates. The expatriate partners were on average 44 years old. Approximately 90% of the partners had relocated with the expatriate and 64% were currently unemployed. The majority of expatriates were French (57%) and the rest of the sample was diverse (e.g. Spain, Sweden, USA, India). No one host country was dominant and host countries were representative of six continents.

Measures

Given our diverse sample, we made the survey available in several languages (i.e. English, Spanish and French). The surveys were translated into Spanish and French from the English version and then back-translated to ensure accuracy.

Partner family role adjustment and expatriate family role adjustment were assessed with a measure from Shaffer et al. (2016). This measure consisted of six items (1 = not at all to 5 = to a great extent). Respondents were asked to rate the extent to which they feel comfortable with aspects of their family life since they, or their partner, became a global employee. A sample item is: “My participation in family activities and tasks” and the internal consistency reliabilities for partner and expatriate family role adjustment were .83 and .88, respectively.

We measured expatriate general self-efficacy with a scale from Chen et al. (2001). The measure consisted of eight items (1 = strongly disagree to 5 = strongly agree). A sample item is: “I will be able to successfully overcome many challenges” and this measure’s reliability was .91.

Expatriates’ family role engagement was measured using a slightly modified scale from Rothbard (2001). This measure consists of nine items (1 = strongly disagree to 7 = strongly agree) and captures both the attention (four items) and absorption (five items) aspects of engagement. Sample items are: “I focus a great deal of attention on my family” (attention) and “When I am with my family, I often lose track of time” (absorption).

Work role engagement was measured with Rothbard’s (2001) scale and consists of nine items, again capturing the attention and absorption aspects of engagement. The items were measured from 1 = strongly disagree to 7 = strongly agree. Sample items are: “I focus a great deal of attention on my work” (attention) and “When I am working, I am completely engrossed by my work” (absorption).

We conducted a confirmatory factor analysis (CFA) to evaluate the structure of the family and work role engagement constructs, which were each conceptualized as consisting of two distinct dimensions: attention and absorption (Rothbard, 2001). We compared a series of measurement models in Mplus 8.2 with robust maximum likelihood estimation starting with a model that combined the absorption and attention dimensions. This two-factor model did not demonstrate an acceptable fit (N = 105; χ² = 728.75, p < .001; df = 134; CFI = .56; RMSEA = .21; SRMR = .15). The four-factor model on the other hand showed a considerable improvement (N = 105; χ² = 327.30, p < .001; df = 129; CFI = .85; RMSEA = .12; SRMR = .10). However, the model still did not fit the data well and there were two items cross-loading on other factors (i.e. “I pay a lot of attention to my family” and “Nothing can distract me when I am taking care of my family”). Removing these cross-loading items avoids introducing bias in the analysis as a result of possible partial overlap between the dimensions (Kline, 2011). Upon removal of these two items, the four-factor model fit the data well (N = 105; χ² = 157.01, p < .001; df = 98; CFI = .95; RMSEA = .08; SRMR = .05) and all loadings were significant and above .60. Thus, we test our hypotheses by looking specifically at the attention and absorption dimensions of family and work engagement. The reliabilities for work role engagement were .90 (attention) and .92 (absorption) and for family role engagement they were .91 (attention) and .93 (absorption).

We considered controlling for several demographic variables. These were gender (0 = male, 1 = female), age (in years), current expatriate assignment duration (in years), number of family members accompanying the expatriate (0 = no relocated family members to 5 = partner and four children), partner current employment (0 = not employed, 1 = employed) and expatriate type (0 = diplomats, 1 = corporate and self-initiated expatriates). In addition, we accounted for expatriate cultural adjustment at Time 1 as a control variable. Cultural adjustment, which represents expatriates’ adjustment to the general conditions in the host-country (Black and Stephens, 1989) is considered to have an association with family role adjustment (Lazarova et al., 2010) and evidence suggests that it also influences the work context (Bhaskar-Shrinivas et al., 2005). Cultural adjustment was measured with an eight-item scale (1 = not at all to 5 = to a great extent) adapted from Black and Stephens’ (1989) measure. The reliability was .87. Finally, we controlled for prior levels of expatriate family and work role engagement by relating them to their respective Time 2 counterparts. Equivalent to the measures at Time 2, the four-factor model for work and family role engagement at Time 1 also had an acceptable fit to the data (N = 105; χ² = 176.80, p < .001; df = 98; CFI = .92; RMSEA = .09; SRMR = .06) and was better than a two-factor model (N = 105; χ² = 394.70, p < .001; df = 103; CFI = .70; RMSEA = .16; SRMR = .11). For family role engagement at Time 1 reliabilities were .83 (attention) and .93 (absorption), and for work role engagement these were .88 (attention) and .85 (absorption).

Analytical procedures

Table 1 presents means, standard deviations, correlations and reliabilities of the study variables. We first examined the correlations between our suggested controls and focal variables. All controls, with the exception of assignment duration and accompanying family members, were significantly correlated with one or more of the focal dependent variables. Consequently, we present our results without these two control variables (Bernerth and Aguinis, 2016). Note that the results remain unchanged when the two controls are included.

OPEN IN VIEWER

Table 1.

Means, standard deviations, internal consistency reliabilities and correlations for study variables.

Variable	Mean	SD	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16	17
1. Expatriate work role engagement (attention) (T2)	6.08	.94	(.90)
2. Expatriate work role engagement (absorption) (T2)	5.05	1.24	.50	(.92)
3. Expatriate family role engagement (attention) (T2)	4.97	1.34	−.04	−.02	(.91)
4. Expatriate family role engagement (absorption) (T2)	4.41	1.46	−.05	.27	.52	(.93)
5. Expatriate family role adjustment	3.59	.86	−.06	−.09	.40	.27	(.88)
6. Partner family role adjustment	3.97	.73	.01	.08	.20	.10	.34	(.83)
7. Expatriate general self-efficacy	4.07	.58	.08	.29	.21	.32	.31	.19	(.91)
8. Expatriate work role engagement (attention) (T1)	6.30	.75	.71	.56	−.03	.09	−.01	.09	.28	(.88)
9. Expatriate work role engagement (absorption) (T1)	4.92	1.18	.34	.71	−.06	.26	−.04	.20	.34	.50	(.85)
10. Expatriate family role engagement (attention) (T1)	4.94	1.26	−.14	−.16	.58	.37	.36	.16	.23	−.04	−.10	(.83)
11. Expatriate family role engagement (absorption) (T1)	4.37	1.52	−.18	.12	.35	.67	.22	.07	.24	−.09	.10	.53	(.93)
12. Expatriate cultural adjustment	3.30	.93	.01	−.16	.19	.05	.35	.06	.03	−.03	−.10	.13	.06	(.87)
13. Partner employment	.36	.48	−.03	.02	.24	.18	−.04	.07	.08	−.05	−.01	.26	.11	.07	−
14. Expatriate gender	.21	.41	.27	.29	−.03	.11	−.11	.07	.11	.30	.33	−.10	.10	−.11	.08	−
15. Expatriate age	44.21	8.91	.14	−.01	−.15	−.28	.10	.04	−.20	.06	−.03	−.19	−.33	.15	−.16	−.23	−
16. Expatriate type	.62	.49	−.02	.02	.04	.24	.14	.01	.14	.01	−.03	.01	.12	.05	−.13	−.42	.09	−
17. Assignment duration a	2.38	2.81	.05	−.02	.06	.09	.19	−.03	.09	−.01	−.01	.01	.06	.30	.14	−.14	.13	.22	−
18. Accompanying family members a	2.01	1.24	−.07	−.09	.07	.09	.12	.11	.07	−.02	−.08	.19	.13	.05	−.02	−.14	−.11	.12	.07

N = 105. All correlations with absolute value larger than .19 are significant at p < .05 and all correlations larger than .25 are significant at p < .01.

Internal consistency reliabilities are in parentheses on the diagonal. Expatriate gender (0 = male, 1 = female), expatriate age (years), partner employment (0 = currently unemployed, 1 = currently employed), expatriate type (0 = diplomats, 1 = corporate and self-initiated expatriates). T1 = Time 1 data collection; T2 = Time 2 data collection.

a

Variables not used in the final analysis.

Given our relatively small sample (N = 105) and the need to test complex moderated mediation models with many variables, we used a series of ordinary least squares regressions (see Tables 2 and 3) in SAS 9.4 to test our hypotheses. To test for moderated mediation (Hypotheses 3 and 5) in an integrative manner, we used the PROCESS macro v.3.2.01 for SAS (Hayes, 2018), which is also based on ordinary least squares regression. The macro uses bootstrapping (k = 10,000) to assess the significance of conditional indirect effects (i.e. moderated mediation) and generates percentile bootstrap confidence intervals. To accurately assess the indirect effects, the process macro also includes the direct relationship from the independent variables to the outcomes. All continuous independent variables were mean centered to aid in interpreting the interaction effects (Aiken and West, 1991).

OPEN IN VIEWER

Table 2.

Ordinary least squares regression (Hypotheses 1 and 2).

Predictor	b	SE	β	p
	Expatriate family role adjustment
Constant	3.55	.17		.000
Expatriate gender	−.11	.21	−.05	.615
Expatriate age	.00	.01	.01	.960
Expatriate type	.17	.17	.09	.336
Expatriate cultural adjustment	.29	.08	.32	.001
Partner employment	−.11	.16	−.06	.490
Partner family role adjustment	.38	.10	.33	.000
	Adj. R2 = .20; F (6, 98) = 5.21 (p = .000)
	Expatriate family role adjustment—H1
Constant	3.72	.16		.000
Expatriate gender	−.23	.20	−.11	.249
Expatriate age	.01	.01	.06	.518
Expatriate type	.01	.17	.00	.995
Expatriate cultural adjustment	.26	.08	.28	.002
Partner employment	−.16	.15	−.09	.289
Partner family role adjustment	.35	.10	.30	.001
Expatriate general self-efficacy	.33	.13	.23	.014
Partner family role adjustment × Expatriate general self-efficacy	−.33	.15	−.19	.033
	Adj. R2 = .28; F (8, 96) = 6.15 (p = .000)
	Expatriate family role engagement (attention) (T2)—H2
Constant	4.75	.23		.000
Expatriate gender	.13	.30	.04	.665
Expatriate age	−.01	.01	−.07	.403
Expatriate type	.11	.24	.04	.647
Expatriate cultural adjustment	.09	.12	.06	.494
Partner employment	.33	.24	.12	.159
Expatriate family role adjustment	.36	.14	.23	.015
Expatriate family role engagement (attention) (T1)	.48	.10	.45	.000
	Adj. R2 = .36; F (7, 97) = 9.31 (p = .000)
	Expatriate family role engagement (absorption) (T2)—H2
Constant	3.74	.23		.000
Expatriate gender	.48	.28	.14	.092
Expatriate age	−.01	.01	−.08	.303
Expatriate type	.69	.24	.23	.004
Expatriate cultural adjustment	−.05	.12	−.03	.693
Partner employment	.39	.22	.13	.074
Expatriate family role adjustment	.26	.13	.15	.051
Expatriate family role engagement (absorption) (T1)	.54	.08	.56	.000
	Adj. R2 = .50; F (7, 97) = 15.58 (p = .000)

N = 105. Table presents unstandardized estimates (b) with their respective standard error (SE), standardized coefficient (β) and p-value (p). T1 = Time 1; T2 = Time 2.

OPEN IN VIEWER

Table 3.

Ordinary least squares regression (Hypothesis 4).

Predictor	b	SE	β	p	b	SE	β	p
	Expatriate work role engagement (T2)
	Attention				Absorption
Constant	5.85	.15		.000	4.92	.21		.000
Expatriate gender	.30	.19	.13	.123	.27	.26	.09	.303
Expatriate age	.01	.01	.10	.199	.01	.01	.07	.358
Expatriate type	.14	.16	.07	.376	.12	.22	.05	.583
Expatriate cultural adjustment	.07	.07	.07	.367	−.16	.10	−.12	.117
Partner employment	.12	.14	.06	.421	.05	.20	.02	.817
Expatriate work role engagement (attention) (T1)	.89	.10	.71	.000
Expatriate work role engagement (absorption) (T1)					.66	.09	.63	.000
Expatriate general self-efficacy	−.14	.12	−.09	.270	.09	.17	.04	.582
Expatriate family role engagement (attention) (T2)	.04	.06	.06	.495	−.02	.08	−.02	.789
Expatriate family role engagement (absorption) (T2)	−.09	.06	−.13	.147	.09	.08	.10	.285
Expatriate family role engagement (attention) (T2) × Expatriate general self-efficacy	.24	.09	.20	.008	−.08	.12	−.05	.499
	Adj. R2 = .52; F (10, 94) = 12.43 (p = .000)				Adj. R2 = .49; F (10, 94) = 11.01 (p = .000)
	Attention				Absorption
Constant	5.84	.15		.000	4.92	.21		.000
Expatriate gender	.33	.19	.14	.090	.25	.26	.08	.328
Expatriate age	.01	.01	.05	.479	.01	.01	.08	.292
Expatriate type	.15	.16	.08	.331	.11	.21	.05	.594
Expatriate cultural adjustment	.06	.07	.06	.376	−.16	.10	−.12	.117
Partner employment	.06	.14	.03	.693	.07	.19	.03	.728
Expatriate work role engagement (attention) (T1)	.89	.09	.71	.000
Expatriate work role engagement (absorption) (T1)					.66	.09	.63	.000
Expatriate general self-efficacy	−.12	.12	−.07	.347	.08	.17	.04	.628
Expatriate family role engagement (attention) (T2)	.05	.06	.08	.354	−.03	.08	−.03	.756
Expatriate family role engagement (absorption) (T2)	−.11	.06	−.17	.066	.09	.08	.11	.256
Expatriate family role engagement (absorption) (T2) × Expatriate general self-efficacy	.22	.07	.22	.002	−.07	.09	−.06	.429
	Adj. R2 = .54; F (10, 94) = 13.06 (p = .000)				Adj. R2 = .49; F (10, 94) = 11.05 (p = .000)

N = 105. Table presents unstandardized estimates (b) with their respective standard error (SE), standardized coefficient (β) and p-value (p). T1 = Time 1; T2 = Time 2.

Theoretical implications

Our study makes several contributions to COR theory, the literatures on crossover and spillover processes and the work–family interface during expatriation. First, by conceiving partner family role adjustment as a contextual resource for the expatriate, we advance COR theory. Except for the newcomer literature (e.g. Saks and Ashforth, 2000), the affective construct of adjustment has received little attention outside the expatriation context. Importantly, while the partner’s family role adjustment may be a typically stable contextual resource, much like marriage more generally (Hobfoll, 2002; Ten Brummelhuis and Bakker, 2012), it becomes more volatile when the family undergoes major role transitions, as in the case of cross-border relocations. The construct of family role adjustment is reflective of the volatility in the family domain owing to a major role transition. However, we contend that adjustment is not only relevant to the expatriation context but captures a unique state during a range of other prominent role transitions, including career moves, changes induced by external shocks like COVID-19 and other relevant life role transitions. It therefore promises to advance our understanding of COR theory explanations of personal and professional change more broadly.

In a related vein, drawing on the idea that resources co-travel in resource caravans (Hobfoll et al., 2018), the expatriation context allows us to study how personal and contextual resources interact and, in fact, may compensate for one another in the resource accumulation process. Specifically, we theorized and empirically demonstrated that the key personal resource of expatriate self-efficacy can compensate for more transient contextual resources, such as partner family role adjustment. This is important because the individual has more control over personal resources compared with contextual resources, especially during periods of role transitions. However, available contextual resources may also compensate for scarce personal resources. In other words, expatriates with low levels of self-efficacy can benefit from their partners’ adjustment to the family role. Indeed, our results suggest that the level of expatriate family role adjustment is very similar (1) at low levels of self-efficacy and high levels of partner family role adjustment and (2) at high levels of self-efficacy and low levels of partner family role adjustment, respectively. While scholars have pointed to the equifinality concept to suggest that individuals can substitute resources to achieve the same goal (e.g. Halbesleben et al., 2014), our findings suggest that such substitution is not limited to versatile and stable resources like personality factors but it also pertains to certain contextual and more volatile resources.

Second, scholars have repeatedly called for examining salient moderators, especially the role of key personal resources (Hobfoll et al., 2018; Ten Brummelhuis and Bakker, 2012), to advance our understanding of the conditions whereby crossover and spillover may more or less likely occur. Although self-efficacy is a prominent concept in social psychology (e.g. Bandura, 1997; Cast and Burke, 2002; Chen et al., 2001; Gist and Mitchell, 1992) and serves as a key resource in COR theory in particular (Hobfoll et al., 2018), it has received little attention as a moderator of crossover and spillover. Our contention that general self-efficacy can compensate for low levels of contextual resources from the family unit to help expatriates adapt psychologically to their family role reinforces a consistent theme in the management literature, namely, the adaptive benefits of general self-efficacy (Schaubroeck et al., 2000; Siu et al., 2007). Further, by focusing on general self-efficacy as a cognitive resource, our study also moves beyond the focus on affective and behavioral mechanisms for how the work and family domains influence each other (Leslie et al., 2019).

Our finding that the crossover between partner and expatriate family role adjustment was not significant at high levels of expatriate general self-efficacy has additional implications for COR and crossover theory, and expatriate research more broadly. Specifically, whereas personal resources such as general self-efficacy allow individuals to sustain work motivation and expend further energy towards an activity or regulate the level of stress they experience, personal resources may also desensitize individuals to other relevant stakeholders. To the extent that family members’ affective states are potential sources of strain for the focal individual, personal resources such as general self-efficacy can serve as buffers (Brockner, 1988; Cast and Burke, 2002). This is consistent with the predominant focus on a stressor–stress–strain model in expatriate research (Black et al., 1991) and the study of negative antecedents to work-related outcomes such as partner maladjustment (see Bhaskar-Shrinivas et al., 2005). However, to the extent that personal resources desensitize individuals to their partners’ feelings and psychological states, it is also possible that the health of the relationship between the couple suffers. Thus, high levels of self-efficacy may function as a double-edge sword in the relationship. While our study did not focus on this potential consequence, we encourage future research to examine this more closely and extend COR theory to consider not only the positive but also the potential negative effects for relevant stakeholders. Similarly, while we zeroed in on the partner’s family role adjustment and conceived it as a contextual resource for the expatriate, future studies should also account for the partner’s role more broadly. For example, studies can examine partners’ other family role experiences, such as their own engagement, but also the couples’ joint family experiences like marital satisfaction. For dual-career couples, it would also be beneficial to examine the expatriate’s influence on the partner’s own work experiences.

In a related vein, by conceiving general self-efficacy as a boundary condition to crossover, we also contribute to expatriate research that has assumed strong crossover in adjustment between the partner and the expatriate to be automatic (Takeuchi, 2010; Takeuchi et al., 2002). Our results suggest that this is not the case under conditions of high self-efficacy. We would speculate that the strength of crossover in adjustment may also depend on other key personal resources that have not been studied much in the expatriate literature thus far—such as emotional intelligence and optimism—or the family unit—such as joint experience with international relocations. Future research may further examine these and other boundary conditions to advance our understanding of crossover and spillover processes during international assignments, and ultimately expatriate effectiveness.

Third, we establish that expatriate family role engagement is an important influence on their work engagement. While previous research has begun to uncover the within-person processes through which work and family life are interlinked (e.g. Ilies et al., 2017), we know little about how engagement in the family unit may affect salient work-related outcomes. Our results contribute to spillover theory and the broader work–family interface literature by integrating two contradictory propositions as to whether individuals’ engagement in family and domestic duties may enrich or deplete their physical, emotional and cognitive reserves, thus either positively or negatively affecting engagement at work (Byrne and Canato, 2017; Leslie et al., 2019). This expands our understanding of the work–family interface because depletion and enrichment arguments tend to be examined separately (Rodríguez-Muñoz et al., 2014; Takeuchi, 2010) or, if considered jointly (Rothbard, 2001), they are not examined along with the antecedents or consequences of engagement in the family and work domains.

Interestingly, we found that the moderating role of general self-efficacy differed regarding the two dimensions of engagement and was more prevalent for the attention dimension compared with the absorption dimension. Whereas attention captures the duration of focus on and mental preoccupation with a given role, absorption concerns the intensity of focus on the work and family roles. As Rothbard (2001) argues, attention can be conceived as a resource-based motivational construct, while absorption reflects a more intrinsic motivational construct. In other words, attention serves as an invisible, cognitive resource that individuals may deploy in different ways. Thus, attention to the family role may signal to the focal individual sufficient cognitive capacity that can be extended to the work role, thereby increasing attention to the work role. Attention is therefore more likely to be influenced by general self-efficacy, which is equally conceived as a cognitive, personal resource (Bandura, 1986). Overall, our results suggest that COR theory arguments may be more adept at explaining the boundary conditions of spillover processes for the attention than for the absorption dimension of engagement.

In contrast to attention, absorption appears to tap into more affective and interest-based facets of engagement (Rothbard, 2001). In our study, we found that spillover between the absorption dimension of family role engagement and the attention dimension of work role engagement was significantly negative at low levels of expatriate general self-efficacy but not significant at high levels of general self-efficacy. This suggests that being absorbed in the family role may deplete an individual’s resources that could otherwise be used for attention to the work role such that it increases positive emotions regarding activities in the family domain. These positive emotions may indicate less availability for work role-related activities, unless the individual can draw on additional personal resources, such as high levels of general self-efficacy.

Finally, we move beyond the predominant focus on expatriate and partner adjustment (Shaffer and Harrison, 2001; Takeuchi, 2010) to consider how and when expatriates’ experiences in the family domain are related to work outcomes. Further, by demonstrating both negative and positive ways in which experiences in the family domain affect the focal expatriate, we integrate the predominant stressor–stress–strain paradigm in expatriate research (e.g. Black et al., 1991) with a more recent focus on the beneficial aspects of global work (e.g. Nurmi and Hinds, 2016).

Practical implications

Our findings have several implications for corporate practice. Given the important role of family role engagement for work role engagement, organizations are well advised to help support and cultivate their expatriates’ family role engagement. Our findings point to at least two ways that organizations can improve expatriates’ family role engagement: organizational support as well as selection and training. First, organizations should give their employees the opportunity to participate more actively in their family roles. For example, organizations may provide more time off, especially at the beginning of the relocation, and avoid overburdening expatriates owing to back-to-back travel demands. They may also offer flexible work practices and flexible office hours to expatriates. Organizations should also provide more support to family members in order to facilitate their international experience. To the extent that partners experience the international relocation as less taxing, they will be more adjusted to their family role, which may help the expatriate to be more adjusted and, ultimately, engaged in their family role.

Second, global mobility professionals should focus on hiring expatriates with high levels of general self-efficacy or providing extra support for those with low levels of self-efficacy. This is particularly important if social norms in the host context differ greatly from the home environment, there is a lack of employment opportunities or the partner will experience substantial changes in family dynamics owing to a sudden increase in the expatriate’s work or travel requirements. Organizations may also provide workshops to increase self-efficacy or teach managers how to develop it. At the same time, such organizational interventions may not necessarily work for those who already have high levels of self-efficacy because they may be less susceptible to external stimuli. If it can be further shown that interventions aiming to increase self-efficacy—or support expatriates more broadly—help trainees with low levels but not those with high levels of self-efficacy, training costs could be saved by only training the former individuals. Training effectiveness may be further improved by identifying what works for those with high levels of self-efficacy. For example, coaching might help with developing tactics to leverage positive experiences in the family domain.

Limitations and future research

Although we proposed and tested a theory-based model and collected data at multiple time points and from multiple sources, our study is not without limitations, which represent avenues for future research. First, our research design precludes us from making causal claims. Despite data from two time points, we are unable to eliminate all possible confounding effects owing to omitted variables, or to firmly establish temporal ordering. While causation is best determined through experiments, such a design was not feasible from a practical standpoint, given the type of variables and intra- and inter-individual mechanisms we examined. Nevertheless, the causal ordering of the variables in our model has been informed by theory.

Second, the number of matched expatriate–partner dyads is limited. Although our sample size is somewhat lower than previous expatriate research that studied matched couples (e.g. Shaffer and Harrison, 1998; Takeuchi et al., 2002), we collected data at two distinct time points to expand on prior work. This likely explains the increased sample attrition. A larger sample would have allowed us to test our conceptual model in a more integrative manner. For example, while we took a directional approach to examine the predictors and moderating conditions affecting expatriate work role engagement, we would welcome research that also captures the bidirectionality inherent in crossover and spillover processes. Similarly, this could involve conceptualizing and measuring not only the role of expatriates’ general self-efficacy but also their partners’ self-efficacy. However, teasing out the predictors, consequences and boundary conditions of crossover and spillover while accounting for their reciprocal nature would entail more sophisticated research designs and data requirements, which are particularly difficult to come by in the expatriate domain (Tharenou, 2015). Further, although our findings did not show any significant effect of expatriates’ accompanying children, scholars have called for more research on understanding how children can influence expatriates’ effectiveness (Dimitrova, 2018) and we would encourage future studies to examine this in greater detail.

Third, while work engagement has been established as a key predictor of work performance (e.g. Christian et al., 2011), we would encourage future studies to explicitly test the work performance implications of our conceptual model. We also acknowledge that expatriate effectiveness is necessarily a multidimensional construct that reaches beyond the individual expatriate and the immediate assignment context (Reiche and Harzing, 2018) and we would invite future studies to expand our findings to other measures of effectiveness.