Organisational identification and decision-making use of decoupled performance measurement systems

Abstract

Based on the social identity approach, this study examines how dual organisational identification of subsidiary managers in multinational corporations (MNCs) influences their use of decoupled performance measurement systems (PMS) for decision-making. We use questionnaires to survey 110 managers at Chinese subsidiaries of MNCs. We find that subsidiary managers’ identification with an organisation, either the subsidiary or the corporation, enhances their evaluation as well as their decision-making use of the PMS initiated by the organisation. Furthermore, having a dominant organisational identification reduces the decision-making use of the PMS associated with the other organisation.

Keywords:

• decision-making
• decoupling
• organisational identification
• multinational corporations
• performance measurement systems

1. Introduction

Network-based multinational corporations (MNCs) can find it challenging to balance the local adaptability of the subsidiary with the legitimacy of the whole MNCs (Grøgaard, 2012; Meyer & Rowan, 1977; Oliver, 1991). A key approach to establishing such a balance and enhancing the collaboration is management control systems (MCS) decoupling (Busco, Giovannoni, & Scapens, 2008; Grøgaard, 2012; Lukka, 2007). As an essential element of MCS, performance measurement systems (PMS) are an important mechanism to integrate diverse entities of MNCs (Busco, Frigo, Giovannoni, Riccaboni, & Scapens, 2006; Coates, Davis, Emmanuel, Longden, & Stacey, 1992; Dossi & Patelli, 2010).

MNCs’ headquarters apply PMS to control business operations in subsidiaries and to align their decision-making with the overall corporate strategy. However, standardised formal PMS designed by headquarters (parent-PMS) may not fully interpret the local context or satisfy local demands. This dilemma leads to decoupled PMS (Cruz, Major, & Scapens, 2009; Dossi & Patelli, 2008; Grøgaard, 2012), which means that a subsidiary independently designs and uses its own PMS (subsidiary-PMS) in addition to using the parent-PMS (Dossi & Patelli, 2008). Busco et al. (2008) argue that decoupled PMS help companies manage the tensions between global uniformity and local adaptation and handle conflicts between centralised and decentralised decision-making. Dossi and Patelli (2008) present the first study to examine the use of decoupled PMS for decision-making. In their study, which uses data from Italian subsidiaries of MNCs, they find that subsidiary business controllers’ decision-making is influenced by both the parent-PMS and the subsidiary-PMS; however, the latter has a greater impact. Building on their work, we aim to identify factors that influence the use of the two competing PMS for decision-making.

Several studies suggest that organisational identification influences control practices by affecting employees’ choices of decision-making alternatives, particularly because people tend to select options that mesh with their identification (Bullis & Tompkins, 1989; Ravishankar & Pan, 2008). Based on these studies, we suggest that subsidiary managers’ organisational identification influences their choice of PMS for decision-making, and they tend to choose the PMS that match their salient organisational identification. Our study investigates the influence of organisational identification on MNC subsidiary managers’ use of decoupled PMS.

We use the social identity approach as the theoretical framework of this study. This approach posits that subsidiary managers possess dual organisational identification (DOI); that is, they identify with both the parent company and the subsidiary (Vora & Kostova, 2007). Further, the subsidiary managers’ identification with the parent company (subsidiary) enhances their evaluation of, commitment to, and support for the parent company (subsidiary) (Ashforth & Mael, 1989). We construct three models based on two characteristics of subsidiary managers’ DOI: correspondence and relative magnitude.

We use a survey to collect data from MNCs’ subsidiaries in the Shanghai region of China. The respondents are managers responsible for subsidiary operation. We have 110 valid responses and these data undergo statistical analysis. Most of our empirical findings follow our predictions and show that subsidiary managers’ identification with the parent company (subsidiary) enhances the evaluation of the parent-PMS (subsidiary-PMS) and stimulates their use for decision-making. The results also suggest that subsidiary managers’ organisational identification influences which PMS they will use for decision-making, and they tend to use the PMS that align with their dominant organisational identification.

This study contributes to the MCS literature in two ways. First, it provides novel evidence that organisational identification directly affects the application of MCS. Consensus is missing in the prior literature on the interaction between organisational identification and the MCS. Some scholars regard organisational identification as a key variable essential for control (Barker & Tompkins, 1994). However, others suggest that only a weak interplay occurs between organisational identification and MCS as they are ‘just informing each other instead of supplementing each other’ (Kärreman & Alvesson, 2004).

Second, the study contributes to the MCS decoupling literature by highlighting the use of decoupled systems and introducing a new perspective on individuals’ impact. Prior literature either describes the phenomenon of MCS (PMS) decoupling or examines the institutional and contextual factors that influence the MCS decoupling process. However, scant research investigates how people use decoupled MCS or examines individuals’ impact on them (Cruz et al., 2009; Friesl & Silberzahn, 2012; Grøgaard, 2012; Lukka, 2007; Markus & Pfeffer, 1983; Meyer & Rowan, 1977; Modell, 2001; Orton & Weick, 1990; Siti-Nabiha & Scapens, 2005). The study is intended to expand the knowledge base by emphasising individuals’ impact on MCS application.

Furthermore, this study underscores the importance of identification management, particularly in MNCs in which employees’ organisational identification tends to be multi-focal. In addition to established practices such as applying advanced management accounting information systems and a modernising organisational structure, headquarters should pay more attention to managing subunit employees’ identification. This identification is essential for MCS application and subsidiary management.

Finally, we contribute to the literature by describing the MCS of MNCs’ subsidiaries in China, one of the most important emerging economies worldwide. In addition, our empirical results serve as a benchmark of adoption of decoupled PMS not only for academics but also for MNCs investing in China.

The rest of the paper is structured as follows. Section 2 reviews the related literature and introduces the theory. Section 3 develops the research hypotheses. Section 4 presents the research method and the data collection procedure. Section 5 focuses on the empirical results. The final section discusses the findings, contributions, and limitations of the study and suggests directions for future research.

2. Literature review

2.1. Decoupled performance measurement systems (PMS)

MNCs with extensive boundaries contend with dual pressure arising from the legitimacy of the whole organisation, on the one hand, and integration with diversified local contexts on the other (Grøgaard, 2012). A corporate strategy designed at the headquarters often conflicts with subsidiaries’ resource-defined business relationship role (Andersson & Forsgren, 1996). To resolve the conflict, sufficient flexibility must be allowed throughout the organisation (Grøgaard, 2012) to support the federative relationship between the parent company and its subsidiaries, which encompasses the disparate goals of the local units and the inclusive goals of the formal organisation (Ghoshal & Bartlett, 1990). Decoupling occurs not only in the strategic dimension but also in the operational dimension (Friesl & Silberzahn, 2012). MNCs usually decouple structural elements from each other and from activities and tend to apply little coordination and control to them (Meyer & Rowan, 1977; Oliver, 1991).

As the key mechanism linking parent companies and their subsidiaries, MCS are often decoupled. Decoupling decreases frictions within the MCS, ensures MCS stability in the formal domain (which confers legitimacy to the organisation), and allows flexibility in the informal domain to adapt to different contexts and needs (Lukka, 2007). According to Grøgaard (2012), MNCs that recognise the importance of both global integration and local adaptability tend to use selective approaches to secure consistent common management control without eliminating subsidiaries’ ability to address local needs. In short, MCS decoupling accommodates diversified local contexts and still meets the requirement for common MCS (Cruz et al., 2009; Grøgaard, 2012; Lukka, 2007).

As a central MCS, PMS play an essential role in subsidiary management (Busco et al., 2006; Coates et al., 1992; Dossi & Patelli, 2010; Mahlendorf, Rehring, Schäffer, & Wyszomirski, 2012). As with decoupled MCS, the use of decoupled PMS arises from the interaction between the parent company and the local subsidiary (Burns & Scapens, 2000; Dossi & Patelli, 2008; Siti-Nabiha & Scapens, 2005), as well as the balance between global unity and local flexibility (Busco et al., 2008). In a study of 15 German, US, and UK MNCs, Coates et al. (1992) suggest different PMS are used in subsidiaries and at headquarters because each organisation has different objectives. The researchers find that subsidiaries adopt a wider range of performance indicators compared with headquarters (Coates et al., 1992). They also address headquarters’ limitations in determining subsidiaries’ PMS due to a lack of understanding and concern about local objectives and needs.

Dossi and Patelli (2008) conceptualise PMS decoupling in MNCs as the subsidiary independently designing and autonomously using its own PMS as well as those that headquarters provide and require. The authors examine the use of the decoupled PMS for decision-making of business controllers in Italian subsidiaries of MNCs and find that 86% of the survey sample use both PMS for decision-making, with the subsidiary PMS having a stronger influence.

2.2. Social identity approach

2.2.1. Social identity theory and self-categorisation theory

This study uses the social identity approach as the theoretic framework. This approach investigates individuals’ cognition and behaviour in social contexts (Haslam, 2004; Haslam, Powell, & Turner, 2000; Hogg & Abrams, 1988; Tajfel, 1974; Van Dick, 2001), and it draws from both social identity and self-categorisation theories.

Social identity is an ‘individual’s self-concept, which derives from his knowledge of his membership of a social group (or groups) together with the emotional significance attached to that membership’ (Tajfel, 1974, p. 69). Social identity theory has three key features. First, the drive to identify with social groups encompasses individuals’ desire for a positive self-image and reduced uncertainty. Thus, individuals tend to join positively evaluated social groups (Tajfel, 1981). Second, individuals construct their self-concept and get to know themselves by identifying with social groups and by aligning with them in the complex social environment (Van Dick, Wagner, Stellmacher, & Christ, 2004; Van Dick, Grojean, Christ, & Wieseke, 2006). Third, social identity clarifies intergroup comparison and behaviour (Tajfel & Turner, 1978).

Turner and his colleagues present self-categorisation theory by suggesting that individuals categorise themselves and others into social groups (Turner, Hogg, Oakes, Reicher, & Wetherell, 1987). Self-categorisation theory focuses on the cognitive basis of individuals’ group behaviour. It suggests depersonalisation, meaning that individuals define themselves by the values, beliefs, and goals of their social group (Turner et al., 1987).

Organisations are one of the most important social groups in modern society, and individuals’ self-image is intimately tied to their organisations. Ashforth and Mael (1989) were the first researchers to apply the social identity approach in an organisational context. Later, they defined organisational identification as the ‘perception of oneness with, or belongingness to, an organisation where the individual defines himself or herself in terms of the organisation in which he or she is a member’ (Mael & Ashforth, 1992, p. 104). Identification with an organisation conveys unity with that organisation as well as separation from others (Albert, Blake, & Dutton, 2000). To achieve a positive self-image, individuals evaluate their organisations with bias but ‘discriminate’ against other organisations, and actively engage in their organisations’ activities to ensure that their performance is superior to that of other organisations. Based on self-categorisation theory (or depersonalisation), organisational identification causes individuals to internalise organisations’ values and goals into their self-concept.

2.2.2. Organisational identification

Organisational identification is often studied in a work-related context, such as its impact on employees’ turnover intentions (Abrams, Ando, & Hinkle, 1998; Dukerich, Golden, & Shortell, 2002; Mael & Ashforth 1995; Olkkonen & Lipponen 2006; Tyler & Blader 2001; van Knippenberg & van Schie 2000); extra-role behaviour (Dukerich et al., 2002; Haslam et al., 2000; Olkkonen & Lipponen, 2006; Tyler & Blader, 2001; Van Dick et al., 2006); in-group cooperative behaviour and out-group competitive behaviour (Dutton, Dukerich, & Harquall, 1994); job satisfaction (van Knippenberg & van Schie, 2000); organisations’ productivity and performance (Haslam, 2004); and post-management of mergers and acquisitions (van Leeuwen, van Knippenberg, & Ellemers, 2003).

2.2.3. Dual organisational identification (DOI) in MNCs

Scholars agree that individuals can simultaneously identify with different organisational levels, such as (1) the personal level with their career; (2) group level with different subunits within the organisation, such as work unit or department; (3) the organisation as a whole; and (4) the occupation or occupational group (Van Dick, 2001). In contrast to earlier studies in which the whole organisation is the only focus of identification, recent studies put greater emphasis on multi-focal organisational identification (Ashforth & Johnson, 2001; Barker & Tompkins, 1994; Johnson et al., 2006; Olkkonen & Lipponen, 2006).

Subsidiary managers’ DOI is an emerging research stream in the multi-focal organisational identification literature (Reade, 2001b, 2001a; Vora & Kostova 2007; Vora, Kostova, & Roth, 2007; Zhang, George, & Chan, 2006). Reade (2001b) was the first to report subsidiary employees’ simultaneous identification with the parent company and the subsidiary. Several empirical studies find that subsidiary employees tend to identify more with the subsidiary than with the parent company (Ashforth & Johnson, 2001; Barker & Tompkins, 1994; van Knippenberg & van Schie, 2000). The employees’ organisational identification is often made salient through daily interaction and frequent communication with the subsidiary and habitual compliance with its routines (Scott & Lane, 2000). Vora and Kostova (2007) propose an elaborate conceptual model on DOI, including its antecedents and its consequences, which offers many potential lines of research. Various antecedents of DOI, such as subsidiary types, prestige and distinctiveness, support from superiors, career opportunity and local staff competency, form the basis for numerous investigations (Reade, 2001b, 2001a; Vora, Kostova, & Roth, 2005; Zhang et al., 2006). However, the consequences of subsidiary managers’ DOI receive little attention with regard to their turnover intentions (Zhang et al., 2006) and their roles in subsidiaries (Vora et al., 2007).

2.2.4. Organisational identification and organisational control

The modern business of management focuses on workers’ ‘insides – the hopes, fears, and aspirations’ rather than directly on their behaviour (Alvesson & Willmott, 2002; Deetz, 1995, p. 87). The significant impact of organisational identification on workers’ beliefs and behaviours makes it an ‘increasingly important modality of organisational control, especially perhaps in larger corporations’ (Alvesson & Willmott, 2002, p. 621). Barker & Tompkins (1994) even regard organisational identification as being more influential than MCS. Wieseke et al. (2012) find that sales representatives’ low organisational identification has a negative impact on sales performance and customer satisfaction. They also find that compensation systems do not offset the negative effects. Despite the recognition of the importance of organisational identification for organisational control, it receives scant mention in the literature (Alvesson & Willmott, 2002). Prior literature shows different opinions about the interaction between organisational identification and MCS. Some scholars suggest that organisational identification enhances employees’ compliance with MCS (Barker & Tompkins, 1994), and some argue that only a weak interplay exists between organisational identification and MCS (Kärreman & Alvesson, 2004).

Bullis and Tompkins (1989) argue that organisational identification affects control practices by influencing selection of decision-making alternatives; specifically, people choose alternatives that accord with their organisational identification. Ravishankar and Pan (2008) find that DOI influences employees’ compliance behaviour. In their study, employees strongly identify with business units and weakly identify with the whole organisation; consequently, they readily comply with business unit initiatives but only loosely comply with organisational initiatives. The two studies show that organisational identification could influence individuals’ choices among available control systems, with people tending to choose the options that align with their salient organisational identification. We subsequently suggest that subsidiary managers’ organisational identification influences their choice of the two competing PMS for decision-making, and they will choose the PMS that matches their organisational identification.

Next, we construct hypotheses and models based on the social identity approach to investigate the relationship between subsidiary managers’ organisational identification and their use of decoupled PMS for decision-making.

3. Hypothesis development

3.1. Correspondence characteristic

The social identity approach suggests that subsidiary managers’ DOI has two characteristics: correspondence and relative magnitude. Based on these characteristics, we establish different models and hypotheses regarding DOI’s impact on the decision-making use of the decoupled PMS.

Self-categorisation theory suggests that individuals can categorise themselves within different levels (Haslam et al., 2000). Each category level represents a distinct identification orientation (Brickson, 2000), and each has its ‘correspondent social motivations, types of significant self-knowledge, and sources of self-worth’ (Brewer & Gardner, 1996, p. 83; Brickson, 2000). Several empirical studies confirm the correspondence between identification foci and their related antecedents and consequences (Christ, Van Dick, Wagner, & Stellmacher, 2003; Haslam et al., 2000; Olkkonen and & Lipponen, 2006; Reade, 2001b; Van Dick et al., 2004).

The correspondence characteristic implies that subsidiary managers’ identification with the parent company affects only the factors pertaining to the parent company (not the subsidiary), and identification with the subsidiary affects only the factors pertaining to the subsidiary (not the parent company). As the parent company designs and provides the parent-PMS, they constitute a parent company factor. Similarly, subsidiary-PMS are a subsidiary factor. Therefore, identification with the parent company (subsidiary) corresponds to the perceived usefulness and the use of parent-PMS (subsidiary-PMS).

Social identity theory’s in-group bias suggests that subsidiary managers’ identification with an organisation enhances their evaluation of, support for, and commitment to that organisation’s activities and systems, including PMS (Ashforth & Mael 1989). Therefore, we suggest that subsidiary managers who identify more closely with the parent company (subsidiary) evaluate the parent-PMS (subsidiary-PMS) more positively and use parent-PMS (subsidiary-PMS) more intensively for decision-making. We consequently propose four hypotheses for correspondence with either the parent or the subsidiary.

Hypothesis 1a: Identification with the parent company is positively associated with the perceived usefulness of the parent-PMS.

Hypothesis 1b: Identification with the subsidiary is positively associated with the perceived usefulness of subsidiary-PMS.

Hypothesis 2a: Identification with the parent company is positively associated with the use of parent-PMS for decision-making.

Hypothesis 2b: Identification with the subsidiary is positively associated with the use of subsidiary-PMS for decision-making.

3.2. Relative magnitude characteristic

The relative magnitude of DOI refers to the relative strength between identification with the parent company and identification with the subsidiary (Vora & Kostova, 2007). As the depersonalisation mechanism (self-categorisation theory) suggests, having a stronger identification with one organisation causes subsidiary managers to incorporate the objectives and value of the organisation in their own self-concept. This in turn enhances their evaluation and support of the organisation and reduces their evaluation and support of other organisations. Vora and Kostova (2007) find that the relative magnitude of DOI influences subsidiary managers’ role behaviour. If subsidiary managers identify more strongly with the parent company than with the subsidiary, they tend to behave as representatives of the parent company, functioning as implementers of its strategy and initiatives. Consequentially, parent-PMS will garner a better evaluation and greater use, while subsidiary-PMS will have a diminished evaluation and lower use for decision-making. However, if subsidiary managers identify more strongly with the subsidiary than with the parent company, they tend to behave as a defender of the subsidiary and are able to interpret local context and to recognise and support local demand. Thus, subsidiary-PMS which address local needs will receive a better evaluation and greater use, while parent-PMS will have a lower evaluation and less use for decision-making. We define the relative magnitude of DOI as the strength of the identification with the subsidiary minus the strength of the identification with the parent company. We suggest that the higher the relative magnitude of DOI, the more the subsidiary managers’ role turns from being frontline implementers to being national advocates and defenders. This leads to four hypotheses.

Hypothesis 3a: The relative magnitude of DOI is inversely associated with the perceived usefulness of parent-PMS.

Hypothesis 3b: The relative magnitude of DOI is positively associated with the perceived usefulness of subsidiary-PMS.

Hypothesis 4a: The relative magnitude of DOI is inversely associated with the use of parent-PMS for decision-making.

Hypothesis 4b: The relative magnitude of DOI is positively associated with the use of subsidiary-PMS for decision-making.

Scholars in other research fields agree on the positive relationship between perceived usefulness and actual use (Davis, 1989; Saadé, 2007), although opinions differ among management accounting scholars. Some suggest that PMS perceived as more useful are more likely to be employed, but others argue that organisational requirements may force individuals to use certain PMS regardless of their perceived usefulness (Chenhall, 2006). Since subsidiaries that use decoupled PMS have autonomy and flexibility to use self-designed subsidiary-PMS for decision-making, we assume that the external pressure to use specific PMS for decision-making is very low. This leads to two hypotheses.

Hypothesis 5a: The perceived usefulness of parent-PMS is positively associated with the use of parent-PMS for decision-making.

Hypothesis 5b: The perceived usefulness of subsidiary-PMS is positively associated with the use of subsidiary-PMS for decision-making.

We construct Model 1 and Model 2 based on the correspondence characteristic and Model 3 according to the relative magnitude characteristic. Figures 1 to 3 summarise the three models and hypotheses.

Figure 1. Variables and hypotheses of Model 1.

4. Research method

4.1. Data collection procedure

We collect data from MNCs’ subsidiaries located in China for two reasons. First, many MNCs have foreign subsidiaries in China; however, the literature on MCS of these subsidiaries is sparse. Therefore, we gather data through a survey and identify benchmarks regarding the management control practices of foreign subsidiaries in China. Second, due to China’s unique cultural and social characteristics as well as its rapidly emerging commercial market, MNCs’ subsidiaries are in an environment that is more complex and greatly different from that of parent companies’ domestic markets (especially for Western companies). MNCs have difficulty in interpreting the local context and being aware of the needs of their Chinese subsidiaries, making the design of suitable PMS challenging for them. We are thus motivated to investigate the application of decoupled PMS in a sample of Chinese subsidiaries.

In order to achieve face validity, clarity, and interpretability (Dillman, Smyth, & Christian, 2008), we conducted a pre-test of the survey with five academics and two practitioners. The seven pre-testers provided feedback and suggestions that helped us modify and finalise the questionnaire. The questionnaire has both English and Chinese versions.¹ We used the back-translation method (Brislin, 1970) to ensure that the content of both versions is the same.

We collected data from foreign subsidiaries in the Shanghai region of China. The companies in our study met two requirements: (1) they are subsidiaries of MNCs and their foreign parent companies own at least a 50% share in them; and (2) the respondents have responsibility for subsidiary operations. We collected data from September 2010 to January 2011 in two batches. The first batch draws on the database of the German Chamber of Commerce in China. We sent emails to 850 subsidiary managers of Germany-based MNCs to invite them to participate in the online survey. We promised participants anonymity and a benchmarking report as an incentive for truthful responses (Dillman et al., 2008). We received 160 responses (response rate: 18.8%). As the research questions require the use of both subsidiary-PMS and parent-PMS in the subsidiary, we excluded 50 responses which do not implement subsidiary-PMS.² We removed an additional 59 responses due to multiple missing values. Finally, 51 responses, which use both parent-PMS and subsidiary-PMS from the first batch, are included in our data analysis.

In parallel, we employed an independent Chinese market research company with experience of conducting similar research projects for data collection (Hoskisson, Eden, Lau, & Wright, 2000; Mahlendorf et al., 2012; Zhang et al., 2006). The research company randomly selected 820 potential firms from their internal database. Research assistants conducted face-to-face interviews at the subsidiary sites. Respondents completed a standardised questionnaire, identical to the online questionnaire, during the interview. The interviews lasted for 60 to 90 minutes each. We received 120 responses (response rate: 14.6%), from which we excluded 45 that do not implement subsidiary-PMS and removed 16 due to multiple missing values. Finally, 59 responses that use both parent-PMS and subsidiary-PMS from the second batch are included in our data analysis.

The response rates of 18.8% and 14.6%, respectively, are reasonable compared with response rates of 12% to 23% in other studies gathering data in China (Chen, Paik, & Park, 2009; Chow, Duh, & Xiao, 2007; Fleming, Chow, & Chen, 2009; Luo & Park, 2001; Mahlendorf et al., 2012). Altogether we have 110 cases, which use both parent-PMS and subsidiary-PMS, for data analysis. We check the two batches to ensure there are no duplicate respondents or companies. We analyse the three models based on the two batches separately and do not find obvious differences between them, which suggests that pooling the data will not have a systematic effect on the final results. Adopting a similar approach, we do not find significant differences between the subsidiaries of Germany-based MNCs (48.2%) and other subsidiaries (51.8%).

4.2. Variable measurement

We mostly use established constructs from the literature to measure the variables, with adaptation of the wording for the current research context, to enhance content validity and facilitate comparability among studies.

We implement Mael and Ashforth’s (1992) measure of organisational identification to assess DOI. Many identification-related studies use this measure (George & Chattopadhyay 2005; Lipponen, Helkama, & Juslin, 2003; Mael & Tetrick, 1992; Olkkonen and & Lipponen, 2006; Vora et al., 2007), and it is valid in a Chinese context (Zobel, 2000). We use two parallel 7-point Likert-type instruments (1 = strongly disagree; 7 = strongly agree) with six items each to measure identification with the parent company (IDPAR) and identification with the subsidiary (IDSUB). We measure the relative magnitude of DOI (RMDOI) by the difference between the total IDSUB score and the total IDPAR score (RMDOI = IDSUB – IDPAR).

We adopt the four most commonly used items of perceived usefulness that Legris (2003) describes to measure the perceived usefulness of parent-PMS (PUPARPMS) and the perceived usefulness of subsidiary-PMS (PUSUBPMS) in two parallel 7-point Likert-type instruments (1 = strongly disagree; 7 = strongly agree), each with four items.

Measurements from Dossi and Patelli (2010) and Henri (2006) are combined and slightly modified to construct two parallel 7-point Likert-type instruments (1 = low usage; 7 = high usage) with 37 items each to measure the decision-making use of parent-PMS and subsidiary-PMS. Furthermore, for both instruments, PMS respondents can mark ‘not included’ if the performance measures do not exist. In this case, the score is zero. The total score of the 37 items of parent-PMS serves as the measurement of the decision-making use of parent-PMS (UPARPMS), while the total score of the parallel 37 items measuring subsidiary-PMS serves as the measurement of the decision-making use of subsidiary-PMS (USUBPMS).

We use slightly different control variables in the three models as they contain different dependent and independent variables. The common control variables in the three models are as follows. A subsidiary’s dependence on its parent company (DEPEN) determines the negotiation power between the two organisations, which influences the subsidiary’s resistance to institutional pressure and its use of decoupled PMS (Kostova & Roth 2002; Oliver 1991; Schotter & Beamish 2011). We refer to the measures of DEPEN that Kostova and Roth (2002) use. Organisational centralisation (CENTR) is one of the most important aspects of organisational structure; it encompasses headquarters’ authority as well as a subsidiary’s flexibility and autonomy (Harzing, 2000). Centralisation increases the intensive use of common MCS and parent-PMS (Coates et al., 1992) and has a positive impact on the subsidiary managers’ identification with the parent company (Vora & Kostova, 2007). We develop measures of CENTR from Gordon and Narayanan (1984) and Harzing (2000). Perceived environment uncertainty (PEU) is included as a control variable because of its impact on organisational structure as well as on the perceived usefulness of PMS (Chenhall & Morris, 1986; Lal & Hassel, 1998) and their actual use (Gordon & Narayanan, 1984; Pedersen & Sudzina, 2012). We use the measures of PEU from Pedersen and Sudzina (2012) as well as Gordon and Narayanan (1984). We also control subsidiary performance (PER), which is an important dimension of subsidiary prestige and thus has a positive influence on subsidiary managers’ organisational identification (Ashforth & Mael, 1989; Reade, 2001b). We adopt the four items of the eight-item scale from Spieker (2004) to measure PER in our study. Subsidiary size (SIZE) is defined by subsidiaries’ annual sales revenue as reported by the respondents to control for size-related effects (Gallo & Christensen, 2011; Haldma & Lääts, 2002). We also control for respondent age (AGE) and job tenure (TENURE) because the older an individual or the longer his or her tenure, the more likely he or she possesses a strong organisational identification (Ashforth & Mael, 1989; Riketta, 2005). TENURE reflects the duration of respondents’ employment at subsidiaries in years. In addition, we use a dummy variable NATION to control for cultural impact (respondents with Chinese nationality = 1, respondents with other nationalities = 0). Tsui (2001) finds that Chinese and Western managers use management accounting systems differently due to their cultural differences. The collectivism of Chinese culture fuels a higher attachment to organisations compared with Western culture (Hofstede, 2007). Hence, subsidiary managers with a Chinese cultural background might have higher organisational identification than those with a Western cultural background. On the other hand, a large power discrepancy in Chinese culture underlies individuals’ strong respect for organisational authority, and individuals with a Chinese cultural background usually take less initiative and are more compliant with rules or systems of a higher authority. Compared with subsidiary managers with a Western cultural background, Chinese subsidiary managers might thus accept parent-PMS more easily and use them more. They also may be less motivated to design subsidiary-PMS and to use them less if they exist. Therefore, we control for respondents’ nationality to eliminate national cultural impact. Furthermore, we control for sample origin by introducing a dummy variable SAMPLE (online questionnaire = 0; paper-based questionnaire = 1).

The involvement of the subsidiary in the design of parent-PMS (IPARPMS) is an additional control variable in Models 2 and 3 because it could increase the use of parent-PMS (Dossi & Patelli, 2008). Figures 1 to 3 depict all the variables.

Figure 2. Variables and hypotheses of Model 2.

Figure 3. Variables and hypotheses of Model 3.

Table 1 includes descriptive statistics of the sample in terms of industries, parent company locations, and respondent nationalities and positions. The sample companies operate in a wide range of industries, with a majority in manufacturing.

Table 1. Sample descriptive statistics.

	Frequency	%		Frequency	%
Panel A. Industry distribution
Aerospace	2	1.8	Food and beverages	1	0.9
Apparel	4	3.6	Industrial and commercial machinery	17	15.5
Automotive	10	9.1	Printing, publishing and media	2	1.8
Business services	9	8.2	Metal producer	3	2.7
Chemical and allied products	11	10.0	Metal product manufacturers	7	6.4
Construction	4	3.6	Oil, gas and coal	1	0.9
Domestic appliances	8	7.3	Retailers	1	0.9
Drugs, cosmetics and healthcare	1	0.9	Logistics	3	2.7
Electrical	2	1.8	Others	11	10.0
Electronics	11	10.0	Total	110	100.0
Panel B. Parent company locations
Hong Kong and Taiwan	5	4.5	UK	8	7.3
Germany	53	48.2	Other European countries	8	7.3
Japan	11	10.0	Other Asian countries	3	2.7
Korea	7	6.4	Others	1	0.9
USA	14	12.7	Total	110	100.0
Panel C. Respondent nationalities
China	70	63.6	Other Asian countries	2	1.8
Germany	28	25.5	Others	1	0.9
USA	2	1.8	Subtotal	108	98.2
UK	2	1.8	Missing values	2	1.8
Other European countries	3	2.7	Total	110	100.0
Panel D. Respondent positions
General manager	25	22.7	Functional manager (e.g. marketing managers, production managers)	27	24.5
CEO	12	10.9	Senior controller	2	1.8
CFO	7	6.4	Senior financial accountant	3	2.7
Chief of controlling department	7	6.4	Others	16	14.5
Chief of accounting/ finance department	11	10.0	Total	110	100.0

In the correlation matrix of all variables in Table 2, we find significant and positive correlations between IDPAR and UPARPMS (r = 0.192, P < 0.05), IDSUB and PUSUBPMS (r = 0.337, P < 0.001), RMDOI and PUSUBPMS (r = 0.207, P < 0.05), PUPARPMS and UPARPMS (r = 0.290, P < 0.001), and PUSUBPMS and USUBPMS (r = 0.229, P < 0.05). These correlations are in line with our hypotheses H2a, H1b, H3b, H5a and H5b. In addition, IDPAR correlates to a minor extent to PUSUBPMS (r = 0.087, ns) and USUBPMS (r = 0.028, ns), and IDSUB correlates on a very small scale to PUPARPMS (r = 0.079, ns) and UPARPMS (r = 0.033, ns). Our analysis supports the correspondence characteristic of DOI in which identification with the parent company (subsidiary) only relates to the factors that pertain to the parent company (subsidiary). Furthermore, IDPAR and IDSUB are significantly positively correlated (r = 0.469, P < 0.001), which aligns with earlier studies’ empirical findings (Olkkonen & Lipponen, 2006; Vora et al., 2007).

Table 2. Correlations, average variances extracted and descriptive statistics (N = 110).

Variable	(1)	(2)	(3)	(4)	(5)	(6)	(7)	(8)	(9)	(10)	(11)	(12)	(13)	(14)	(15)	(16)	(17)
(1) IDPAR	0.56
(2) IDSUB	0.47^c	0.60
(3) RMDOI	–0.61^c	0.41^c	0.57
(4) PUPARPMS	0.11	0.08	0.06	0.86
(5) PUSUBPMS	0.09	0.34^c	0.21^b	0.12	0.83
(6) UPARPMS	0.19^a	0.03	–0.16	0.29^c	0.080	–
(7) USUBPMS	0.03	–0.01	–0.05	0.21^a	0.23^b	0.47^c	–
(8) DEPEN	0.05	–0.28^c	–0.30^c	0.15	–0.02	0.15	0.31^c	0.53
(9) PEU	0.17	0.16	–0.04	–0.14	0.05	–0.08	–0.24^b	–0.22^b	0.41
(10) PER	0.01	0.14	0.12	0.14	0.07	0.34^c	0.15	–0.01	–0.41^c	0.57
(11) CENTR	0.12	0.30^c	0.14	–0.02	0.09	0.05	0.13	–0.04	0.00	–0.05	0.57
(12) AGE	0.19^a	0.25^c	0.02	–0.18	0.01	0.01	0.09	–0.08	0.17	–0.14	0.13	–
(13) IPARPMS	0.07	0.03	–0.05	0.34^c	0.06	–0.11	–0.10	–0.19^a	0.04	0.12	–0.03	–0.17	–
(14) SIZE	–0.05	–0.13	–0.06	0.13	–0.11	–0.06	0.05	0.10	–0.13	0.03	–0.10	–0.01	–0.12	–
(15) TENURE	0.01	0.00	–0.01	–0.32^c	–0.08	–0.06	–0.05	–0.19^a	0.08	0.01	0.06	0.40^c	–0.22^b	–0.00	–
(16) NATION	–0.33^c	–0.43^c	–0.02	–0.02	–0.09	0.14	0.08	0.22^a	–0.40^c	0.28^c	–0.07	–0.42^c	–0.14	–0.32^c	–0.09	–
(17) SAMPLE	–0.23^b	–0.38^c	–0.08	0.08	–0.02	0.14	0.15	0.23^a	–0.41^c	0.22^b	0.34^c	–0.40^c	–0.16	–0.21^b	–0.08	0.68^c	–
MIN (actual)	1.00	1.00	3.00	1.00	2.00	11.00	22.00	1.00	1.00	2.00	1.00	25.00	1.00	0.06	1.00	0.00	0.00
MAX (actual)	7.00	7.00	13.00	7.00	7.00	222.00	232.00	7.00	7.00	7.00	7.00	61.00	7.00	40,000.00	17.00	1.00	1.00
MEAN	4.98	5.67	7.70	4.45	4.51	112.03	142.58	2.97	2.76	5.35	4.65	1.49	3.90	1188.09	5.34	0.68	0.50
SD	1.55	1.30	1.49	1.50	4.66	54.84	43.81	3.61	42.22	1.16	1.13	1.50	2.12	4070.40	3.49	0.47	0.50
Cronbach	0.84	0.86	0.85	0.95	0.93	–	–	0.83	0.76	0.75	0.76	–	–	–	–	–	–
Composite reliability	0.88	0.90	0.89	0.96	0.95	–	–	0.87	0.83	0.84	0.84	–	–	–	–	–	–

Significance (two-tailed test):

^a shows p < 0.1,

^b shows p < 0.05,

^c shows p < 0.01.

5. Results

5.1. Measurement model

We evaluate factor loadings to check whether single items can reasonably represent their respective latent variables. All factor loadings are above the critical level of 0.50 that Hair & Anderson (2010) suggest. We evaluate the reliability of the latent variables by Cronbach’s α and composite reliability. As Table 2 shows, all measurement scales exhibit satisfactory levels of Cronbach’s α, ranging from 0.75 to 0.95, and all are above the cut-off of 0.70 (Hair & Anderson 2010). The values of composite reliability range from 0.83 to 0.96 and surpass the limit of 0.60 (Bagozzi & Yi, 1988). The values of average variance extracted from all the latent variables, except for one, are higher than the cut-off value of 0.50 (Bagozzi & Yi, 1988; Fornell & Larcker, 1981). Perceived environment uncertainty (PEU) is 0.41 and only falls slightly below the threshold. Thus, taken together, these results support convergent validity of all latent variables. Furthermore, we test the discriminant validity by using two procedures from Fornell and Larcker (1981). First, nearly all constructs extract a higher average variance than the variance they share with other factors. The only exception is identification with the parent company (IDPAR), which extracts a lower average variance than the variance it shares with the relative magnitude of DOI (RMDOI). We think the exception arises from the mathematical relation between the two constructs (RMDOI = IDSUB − IDPAR). Second, all measurement items have higher loadings on their assigned factor than on all other factors. We conduct Harman’s one-factor test to evaluate the presence of common method bias (Harman, 1967). The basic assumption of the test is that a single factor will emerge from the factor analysis if a significant amount of common method variance is present (Podsakoff & Organ, 1986). The results indicate that the first factor only explains 20.84% of all variance, which means that no single factor explains the majority of the variance of all variables. Thus, common method bias does not seem to be an issue.

5.2. Structural model and tests of hypotheses

We rely on the structural equation modelling (SEM) technique, which is based on partial least squares (PLS). We use Smart PLS 2.0 M3 inherent bootstrapping as a resampling technique (500 random samples) to assess the properties of the latent variable derived from the survey instruments, to evaluate the structural model, and to test the postulated hypotheses (Ringle & Spreen, 2007). PLS-SEM maximises the explained variance of the dependent constructs. Further, the method’s flexibility and its high statistical power make it well-suited for applications that aim at prediction or theory building (Hair, Ringle, & Sarstedt, 2011), and PLS procedures are especially suitable for small sample size (Henseler, Ringle, & Sinkovics, 2009; Sosik, Kahai, & Piovoso, 2009). These features of PLS-SEM accord well with this study, which is exploratory and has a small sample size. The PLS-SEM assessment is a two-step procedure comprising separate assessments of the measurement model, which describes the relations between item measurement (survey responses) and the latent constructs they capture, and the structural model, which states the relationships between latent constructs. We rely on the path coefficients’ magnitude and sign and their statistical significance to assess different models. Figures depict the results of the structural models, and Table 3 summarises the related statistics results.

Table 3. Standardised PLS parameter estimates of structural models.

Model 1 - Parent company level hypotheses	Path coefficient	T statistics
Identification with the parent company → Decision-making use of parent-PMS	0.187*	1.93
Identification with the parent company → Perceived usefulness of parent-PMS	0.104	0.74
Perceived usefulness of parent-PMS → Decision-making use of parent-PMS	0.332***	3.69
Controls
Subsidiary's dependence on parent company → Perceived usefulness of parent-PMS	0.144	1.12
Subsidiary's dependence on parent company → Decision-making use of parent-PMS	0.029	0.31
Subsidiary performance → Decision-making use of parent-PMS	0.295***	2.61
Perceived environment uncertainty → Decision-making use of parent-PMS	0.119	0.94
Centralisation → Decision-making use of parent-PMS	0.060	0.45
Age → Decision-making use of parent-PMS	0.115	0.76
Local involvement in the design of parent-PMS → Decision-making use of parent-PMS	–0.225**	2.04
Size → Decision-making use of parent-PMS	–0.095	0.56
Job tenure → Decision-making use of parent-PMS	–0.039	0.33
Nationalities → Decision-making use of parent-PMS	0.130	0.92
Sample → Decision-making use of parent-PMS	0.013	0.10
Model 2 - Subsidiary level hypotheses	Path coefficient	T statistics
Identification with the subsidiary → Decision-making use of subsidiary-PMS	–0.012	0.09
Identification with the subsidiary → Perceived usefulness of subsidiary-PMS	0.375***	4.64
Perceived usefulness of subsidiary-PMS → Decision-making use of subsidiary-PMS	0.206**	2.10
Indirect effect
Identification with the subsidiary → Perceived usefulness of subsidiary-PMS → Decision-making use of subsidiary-PMS	0.077*	1.65
Controls
Subsidiary's dependence on parent company → Decision-making use of subsidiary-PMS	0.217*	1.86
Subsidiary's dependence on parent company → Perceived usefulness of subsidiary-PMS	0.098	0.91
Subsidiary performance → Decision-making use of subsidiary-PMS	–0.037	0.29
Perceived environment uncertainty → Decision-making use of subsidiary-PMS	–0.172	1.44
Centralisation → Decision-making use of subsidiary-PMS	0.232*	1.78
Age → Decision-making use of subsidiary-PMS	0.166	1.57
Size → Decision-making use of subsidiary-PMS	–0.296*	1.90
Job tenure Decision-making use of subsidiary-PMS	–0.038	0.41
Nationalities → Decision-making use of subsidiary-PMS	0.129	0.91
Sample → Decision-making use of subsidiary-PMS	0.064	0.53
Model 3 - Hypotheses	Path coefficient	T statistics
Relative magnitude of dual organisational identification → Decision-making use of parent-PMS	–0.195**	2.41
Relative magnitude of dual organisational identification → Decision-making use of subsidiary-PMS	–0.059	0.68
Relative magnitude of dual organisational identification → Perceived usefulness of parent-PMS	–0.022	0.15
Relative magnitude of dual organisational identification → Perceived usefulness of subsidiary-PMS	0.251**	2.07
Perceived usefulness of parent-PMS → Decision-making use of parent-PMS	0.347***	3.63
Perceived usefulness of subsidiary-PMS → Decision-making use of subsidiary-PMS	0.218**	2.53
Controls
Subsidiary's dependence on parent company → Perceived usefulness of parent-PMS	0.146	1.24
Subsidiary's dependence on parent company → Perceived usefulness of subsidiary-PMS	0.075	0.72
Subsidiary's dependence on parent company → Decision-making use of parent-PMS	–0.022	0.15
Subsidiary's dependence on parent company → Decision-making use of subsidiary-PMS	0.194*	1.68
Perceived environment uncertainty → Decision-making use of parent-PMS	0.121	1.00
Perceived environment uncertainty → Decision-making use of subsidiary-PMS	–0.180**	2.07
Subsidiary performance → Decision-making use of parent-PMS	0.343***	3.06
Subsidiary performance → Decision-making use of subsidiary-PMS	–0.040	0.31
Centralisation → Decision-making use of parent-PMS	0.068	0.53
Centralisation → Decision-making use of subsidiary-PMS	0.239*	1.79
Age → Decision-making use of parent-PMS	0.136	0.93
Age → Decision-making use of subsidiary-PMS	0.168**	1.66
Local involvement in the design of parent-PMS → Decision-making use of parent-PMS	–0.252**	2.29
Size → Decision-making use of parent-PMS	–0.084	0.52
Size → Decision-making use of subsidiary-PMS	–0.294*	1.76
Job tenure → Decision-making use of parent-PMS	–0.061	0.50
Job tenure → Decision-making use of subsidiary-PMS	–0.041	0.42
Nationalities → Decision-making use of parent-PMS	0.085	0.69
Nationalities → Decision-making use of subsidiary-PMS	–0.118	0.83

Significance (two-tailed test, based on bootstrapping with 500 replacement):

^* p < 0.1,

^** p < 0.05,

^*** p < 0.01.

Figure 4. Structural model results of Model 1.

Significance (two-tailed test): * p < 0.1, ** p < 0.05, *** p < 0.01.

Figure 5. Structural model results of Model 2.

Significance (two-tailed test): * p < 0.1, ** p < 0.05, *** p < 0.01.

Figure 6. Structural model results of Model 3.

Significance (two-tailed test): * p < 0.1, ** p < 0.05, *** p < 0.01.

In Model 1, the R² value of the decision-making use of parent-PMS is 0.303. In Model 2, the R² value of the decision-making use of subsidiary-PMS is 0.347. In Model 3, the R² value of the decision-making use of parent-PMS is 0.306, and the R² of the decision-making use of subsidiary-PMS is 0.350. The three models show medium to high effect size (Cohen, 1988).

We measure the global fit of the three models (Tenenhaus et al., 2005). The goodness-of-fit (GoF) is 0.301, 0.359, and 0.353 in Model 1, Model 2, and Model 3, respectively. According to the baseline values that indicate 0.1 = small GoF, 0.25 = medium GoF, and 0.36 = large GoF (Wetzels, Odekerken-Schroder, & Van Oppen, 2009), all three models indicate good fit.

The results support the hypotheses H1b, H2a, H3b, H4a, H5a, and H5b. The positive and significant path coefficient between identification with subsidiary and the perceived usefulness of subsidiary-PMS (0.375; P < 0.001) supports that if subsidiary managers identify more with their subsidiaries, they tend to assess subsidiary-PMS as being more useful. The result also supports the assumption that identification with the parent company increases the decision-making use of parent-PMS (0.187; P < 0.1). The relative magnitude of DOI increases the perceived usefulness of subsidiary-PMS (0.251; P < 0.05) and reduces the decision-making use of parent-PMS (−0.195; P < 0.05). Model 1 (0.332; P < 0.001) and Model 3 (0.347; P < 0.001) support the positive relationship between the perceived usefulness of parent-PMS on the decision-making use of parent-PMS. Similarly, Model 2 (0.206; P < 0.05) and Model 3 (0.218; P < 0.05) support the positive relationship between the perceived usefulness of subsidiary-PMS and the decision-making use of subsidiary-PMS. Furthermore, we find that identification with the subsidiary indirectly increases the decision-making use of subsidiary-PMS mediated by the perceived usefulness of subsidiary-PMS (0.077, P < 0.1).

We also find that 90 of the 110 subsidiary managers have greater identification with their subsidiary than with their parent company. This finding accords with several empirical studies showing that identification with smaller subunits is stronger than identification with broader units (Ashforth & Johnson, 2001; Barker & Tompkins, 1994; van Knippenberg & van Schie, 2000).

Unexpectedly, the results do not support hypothesis H2b or H4b. The path coefficient between identification with the subsidiary and the decision-making use of subsidiary-PMS is −0.012 (T = 0.09). The path coefficient between the relative magnitude of DOI and the decision-making use of subsidiary-PMS is −0.030 (T = 0.68).

The relative power between the parent company and the subsidiary could explain these unexpected results. The impact of the relative power on subsidiary managers’ identification with the subsidiary is the opposite of its impact on the decision-making use of subsidiary-PMS. According to the social identity approach, if the subsidiary holds stronger power than the parent company, subsidiary managers, to assure a positive self-image, tend to identify more with the subsidiary than with the parent company. Thereby, the relative magnitude of DOI increases. So a subsidiary’s relative power should have a positive impact on identification with the subsidiary as well as on the relative magnitude of DOI. On the other hand, by holding more power than its parent company, a subsidiary is well positioned to negotiate with the parent company for more adaptive parent-PMS (Schotter & Beamish, 2011), which reduces the need to design and use subsidiary-PMS. So, a subsidiary’s relative power may have a negative impact on the decision-making use of subsidiary-PMS.

In short, a subsidiary with stronger relative power has greater involvement in and influence on headquarters’ design of MCS. Consequently, parent-PMS better reflect subsidiaries’ wants and needs and thus the demand for using self-designed subsidiary-PMS drops.

There are various factors which influence the relative power between a subsidiary and its parent company, including a subsidiary’s dependence on a parent company (Pfeffer & Salancik, 1978; Schotter & Beamish, 2011). In this study we find that the correlation between a subsidiary’s dependence on its parent company and identification with the subsidiary is significantly negative (−0.279; P < 0.01), as is the correlation between a subsidiary’s dependence on the parent company and the relative magnitude of DOI (−0.299; P < 0.01). These results are in accordance with Kostova and Roth’s (2002) study, which finds that the less dependent a subsidiary is on its parent company, the more subsidiary managers identify with the subsidiary and the less they identify with the parent company. In addition, we find a significantly positive correlation between a subsidiary’s dependence on the parent company and the decision-making use of subsidiary-PMS (0.313; P < 0.01). These results support our previous assumptions that the relative power between a subsidiary and the parent company has an opposite impact on subsidiary managers’ identification with the subsidiary and on their use of subsidiary-PMS.

However, we do not control for other major factors that influence the relative power; for example, the centrality of the subsidiary and that of the parent company in the network (Ghoshal & Bartlett, 1990). The absence of these control variables is one potential explanation for the non-significant findings regarding some hypotheses.

The results also do not support hypotheses H1a or H3a. The path coefficient between identification with the parent company and the perceived usefulness of parent-PMS is 0.104 (T = 0.74). The path coefficient between the relative magnitude of DOI and the perceived usefulness of parent-PMS is −0.022 (T = 0.15). The results show that identification with the parent company as well as the relative magnitude of DOI has a direct impact on subsidiary managers’ decision-making use of parent-PMS (behaviour) instead of influencing their belief or evaluation regarding parent-PMS. Our findings suggest that organisational identification could have different functions in the evaluation and behaviour dimensions. Organisational identification may not influence subsidiary managers’ evaluation of MCS but it may have a direct impact on their application of MCS.

The dummy control variable NATION has a significant negative relation with identification with a parent company (−0.331, p < 0.001) as well as with identification with a subsidiary (−0.430, p < 0.001). Respondents with Chinese nationality have a significantly lower identification with both the parent company and the subsidiary compared with other respondents who are mostly from Europe (87.5% of the non-Chinese respondents are from Europe). This finding confounds our assumption that Chinese individuals have a naturally strong attachment to social groups such as organisations. Two possible explanations exist for the finding. First, some scholars indicate that the Chinese have high collectivism only in a family setting; at work, they are very individualistic (Wong, 2001; Yeh, 1988). Second, the non-Chinese respondents are mostly expats that the headquarters send to the subsidiary. They have often worked in the headquarters’ organisational hierarchy before and have frequent communication with it now. Besides, they usually have a higher position and assume more responsibility in the subsidiary. Therefore, the non-Chinese respondents probably have higher identification with both the parent company and the subsidiary (Reade, 2001b).

Furthermore, we notice that Chinese subsidiaries of MNCs have much lower adoption of self-designed subsidiary-PMS in comparison with Dossi and Patelli’s (2008) study. As we previously mention, 95 cases are excluded from data analysis because they do not use subsidiary-PMS. Given that 110 subsidiaries do use both parent-PMS and subsidiary-PMS, the adoption rate of the decoupled PMS is 54%, which is much lower than the 89% adoption rate in the Italian subsidiaries of MNCs (Dossi and & Patelli, 2008). The result runs counter to the expectation that Chinese subsidiaries should use more self-designed PMS as the Chinese market is new, emerging, volatile, and very competitive, and the Italian market is rather stable and mature. However, it supports our expectation that the high power discrepancy present in Chinese culture might lead to passivity within Chinese subsidiaries with regard to accepting and using the PMS provided by headquarters as well as a lower inclination to initiate their own (Min, 2007). Furthermore, we think a lack of training and knowledge about modern management accounting in China might socially limit subsidiaries’ capabilities to design their own PMS or even to recognise the shortcomings of the parent-PMS (Chow et al., 2007; Islam & Kantor, 2005).

In order to further analyse the correspondence characteristic of DOI, we construct Model 4 to investigate the relationship between identification with the parent company and the perceived usefulness of subsidiary-PMS and their use in decision-making and Model 5 to test the relationship between identification with the subsidiary and perceived usefulness of parent-PMS and their use in decision-making (see Figures 7 and 8). As the results show, no hypothesis is supported, thus providing evidence for the correspondence characteristic of DOI (i.e., identification with one organisation influences only the factors that pertain to this organisation).

Figure 7. Structural model results of Model 4.

Significance (two-tailed test): * p < 0.1, ** p < 0.05, *** p < 0.01.

Figure 8. Structural model results of Model 5.

Significance (two-tailed test): * p < 0.1, ** p < 0.05, *** p < 0.01.

The results of the three models using the correspondence characteristic and the relative magnitude characteristic are similar and comparable. This consistency further supports the reliability of this study.

6. Discussion and conclusion

This study applies a social identity approach, relying on DOI’s correspondence characteristic and relative magnitude characteristic, respectively, to examine the impact of subsidiary managers’ DOI – identification with the parent company and identification with a subsidiary – on their use of the decoupled PMS (parent-PMS and subsidiary-PMS) for decision-making. The study illustrates the impact of organisational identification on subsidiary managers’ choice of competing PMS for decision-making. Organisational identification can positively influence the evaluation of PMS and place positive and direct impact on the use of PMS. Furthermore, it confirms that given the decoupled MCS, subsidiary managers tend to make decisions by using the MCS which match their dominant organisational identification, while relying less on the other MCS.

This study broadens the MCS decoupling literature by demonstrating how employees’ psychological attachment to an organisation influences their choice of MCS for decision-making. Most prior literature focuses on institutional and contextual factors that trigger or alter MCS decoupling. Few studies centre on individual factors or the application of the decoupled MCS. However, the function of MCS in a business not only depends on the organisational structure or environmental context but also on the employees who implement MCS. Moreover, it is especially important to examine decoupled MCS with regard to individual factors’ impact because the organisational flexibility and adaptability that a decoupled system provides require intensive subjective evaluation and individual decision for choosing the appropriate MCS.

The study also sheds light on the relationship between perceived usefulness and actual use. As the study results show, for both parent-PMS and subsidiary-PMS the perceived usefulness has a positive impact on actual use. We provide evidence that in the context of less external pressure, such as using decoupled MCS, a positive evaluation of the MCS increases the actual use of MCS.

The study reveals a much lower adoption rate of self-designed subsidiary-PMS in foreign subsidiaries located in China compared with those located in Italy (Dossi and & Patelli, 2008). Both cultural (high power discrepancy) and social (lack of training and knowledge) factors could lead to the use of improper MCS in the foreign subsidiaries in China by over-reliance on MCS designed by headquarters that do not address local needs or fulfil subsidiaries’ own strategic goals. We suggest future studies examine the fit between the management control practices of foreign subsidiaries in China and their actual needs. We also advise MNCs to increase their Chinese subsidiaries’ involvement in designing MCS and to offer more internal training or knowledge exchange on MCS design and use for their employees in China.

In addition, we urge MNCs (as well as large organisations) to monitor employees’ organisational identification. Organisational identification in these companies exerts an extraordinary impact on the application of MCS, and the companies’ often decoupled MCS provide individuals latitude to choose between the two MCS. Furthermore, due to companies’ complex organisational structures, multiple hierarchies, and break-ups between different organisational units, employees tend to have multi-focal organisational identification. MNCs thus ought to make more effort to observe and maintain their subsidiary employees’ multiple identifications to achieve proper decision-making for sustainable performance of the organisation as a whole.

This study has some limitations. First, the survey sample is entirely from the Shanghai region, which might decrease the generalisability of the results. Second, we measure the decision-making use of parent-PMS (subsidiary-PMS) by calculating the total score of all 37 performance measures in the questionnaire. Despite the questionnaire, including a broad range of performance measures, it probably does not cover all measures that the sample companies use. Future studies could search for a better measurement of (decision-making) use of PMS. Third, we do not fully control subsidiaries’ relative power in the study, which could explain the two unsupported hypothesis. However, the study reveals interesting relationships among relative power, organisational identification, and MCS application. Future studies could investigate them in depth.

Acknowledgements

The authors are grateful for the insightful comments of the anonymous referees and editors. We thank Xuyi Wang for his great help in the data collection process and for pre-testing the survey. We thank Matthias D. Mahlendorf for his valuable feedback. In addition, we gratefully acknowledge the comments and suggestions by conference participants at the 2012 Manufacturing Accounting Research Conference (June 2012), the Second Annual Conference of China Journal of Accounting Studies (May 2013) and the Research Conference on Management Accounting from the Field (June 2013).

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. A copy of the questionnaire is available from the corresponding author on request.

2. A filter question is incorporated in the questionnaire. Respondents are required to answer whether both parent-PMS and subsidiary-PMS are used in the subsidiary or only parent-PMS.