ادغام سیستم های مالی و حسابداری مدیریت: تاثیر واسطه ای زبان مالی سازگار بر اثر کنترل شیپ

Two fundamental options exist for management accounting system (MAS) design: Either financial records can be used as a database for management accounting (integrated accounting system design), or the MAS can be based upon a separate system, i.e., a third set of books beside financial and tax accounting records. Since the 1990s, many German-speaking firms have changed from the second to the first option, which has instigated a highly controversial debate. Our paper contributes to this debate by empirically analyzing (1) whether the integration of financial and management accounting has a positive impact on controllership effectiveness, and (2) what causal inferences relate both variables. We use structural equation modelling for a sample of 149 dyads surveyed from German top 1500 firms. We identify no significant effect of the technical aspects of MAS integration, but a fully mediating influence of a consistent financial language on controllership effectiveness. Our results thus imply that consistency with financial reporting is an important property of MAS design from management's point of view.

Even though the strategic as well as operational impact of managerial action is directed at non-financial goals, like productivity level, sales volume, market share or stakeholders’ potential for consumption, most firms use accounting information for decision-making and control purposes. Literature on management control systems provides several arguments for this prevalence (Merchant and Van der Stede, 2007 and Malina and Selto, 2004). For example, accounting data are congruent with the organizational goal of profit maximization and can be tailored to managerial decision-making and control problems on each hierarchy level. The accounting model of a firm's transactions is to a large extent causally related to the business's economic success, so that accounting information supports result controls. The predominant use of accounting information as a control mechanism is reflected in management's bonus contracts, which usually include at least one accounting-based measure, e.g., residual income or ratio measures (Murphy, 1999 and Ittner et al., 1997). To provide accounting information for decision-making and control purposes, two fundamental options exist. • On the one hand, the financial accounting records can be used as the main database for management accounting techniques (e.g., product costing or budgeting), reporting and performance measurement. We refer to such a design of the management accounting system (MAS), which is typically observed in Anglo-American firms, as ‘integrated’. Two major advantages can be found with an integrated accounting system design. First, management accounting information is provided at low incremental cost. Second, internal and financial performance measures are easily reconciled on all hierarchy levels, providing management as well as investors with ‘one version of the truth’. This point is of special importance in capital market-oriented firms in which clear links between investors’ targets and management accounting information are needed. Nevertheless, the financial accounting data may not in all cases be suitable for management control purposes, as the underlying accounting regulation is not designed for internal decision-making and/or decision-influencing purposes in the first instance (e.g., Kaplan, 1984). • On the other hand, the MAS can be based upon a so-called separate third set of books beside the financial and tax accounting records. Such a ‘separate’ or ‘dual’ design (Jones and Luther, 2005) has traditionally been used in continental European and especially in German-speaking countries. An integral feature of a separate MAS design is the use of non-GAAP-based accruals for internal planning, budgeting and performance measurement. Such accruals may be imputed costs (e.g., depreciation or cost of material based on replacement values, lump-sum risk provisioning, or opportunity costs for owners’ assets, capital and/or labour input). A major argument supporting a separate accounting system design is the high degree of case-by-case flexibility in measuring resource consumption and output with respect to decision-making and control problems at hand – a philosophy that can be described as ‘different costs for different purposes’. Jones and Luther (2005) claim that the benefits of being able to freely design financial controls under a separate accounting system design might even outweigh the disadvantage of not being able to reconcile internal and external performance measures at top-management or business-segment level. Additionally, the use of cost-based operational controls allows for tight management control structures as the local cost data are aggregated and monitored by top management. In integrated accounting systems, however, local controls are based on non-financial information, with middle management ‘shielding’ local managers from corporate financial goals (Euske et al., 1993). Since the 1990s, an increasing number of German firms have changed their accounting systems from a separate to an integrated design for decision-making and control purposes (Ewert and Wagenhofer, 2007, Wagenhofer, 2006 and Jones and Luther, 2005). Today, professional practice uses pure types of integrated or separated accounting systems as well as hybrid forms (‘partial integration’; Angelkort et al., 2009). In the latter case, the integration of financial and management accounting information is restricted, e.g., to the top hierarchy levels or to selected parts of the financial records’ database. The first German company that openly challenged the tradition of using separate accounting systems was the top 30 technology firm Siemens in 1992/1993 (Ziegler, 1994). Referring to the need for a consistent accounting language for internal as well as external communication purposes, Siemens based its top-management control procedures on the financial accounting database. Ziegler's (1994) seminal paper has thus started a highly controversial debate in the mainly analytical and/or normative streams of German management accounting literature on whether this change has any effects on management decision-making and control, and if so, whether this change is for the better (e.g., Schweitzer and Ziolkowski, 1999 and Schildbach and Wagner, 1995). Our paper aims at adding some insights regarding the debate on separate versus integrated MAS design from an empirical perspective, namely by analyzing the impact of an increased level of integration of accounting systems on controllership effectiveness. Until now, empirical research on this subject – mainly published in German language – has focused either on describing changes in management accounting practice due to the spread of integrated MAS design in German-speaking countries (e.g., Müller, 2006, Wagenhofer and Engelbrechtsmüller, 2006 and Horváth and Arnaout, 1997) or on discussing the role that IFRS play as a contextual factor (e.g., Jones and Luther, 2005 and Weißenberger et al., 2004). Our paper expands this existing body of research by addressing the following research questions: 1. Does an increased level of MAS integration have a positive impact on controllership effectiveness? 2. If so, does the underlying causal inference relate both variables directly, i.e., in a technical fashion, or are they related indirectly via properties of accounting as a financial language for business communication? Our research design is distinctive for several reasons. First, in our study we explicitly link MAS design to controllership effectiveness, which has not been attempted before. Second, we use a dyadic research design surveying both controllers and representatives of general management. This allows us to include management's perspective into our analysis in a valid fashion. Third, we do not restrict our analysis to technical features of the MAS design, but include management's evaluation of the accounting information provided as a financial language, which relates to the construct of conceptual information use (Menon and Varadarajan, 1992, Beyer and Trice, 1982 and Burchell et al., 1980). With respect to the latter point, our research follows the suggestions of Hopwood (1974), who points out that “… accounting is about human behaviour, and its social and behavioural aspects are just as much an indispensable part of the whole as the more traditional technical aspects.” (p. 14) Though identified in one particular national context, the issues raised in our paper relate to the international discussion on management accounting and control and are therefore of interest outside German-speaking countries as well. For example, multinational corporations are confronted with a similar problem regarding their accounting system design if the headquarters operates under a financial accounting regime that is different from the one governing some of the firm's foreign subsidiaries. This may be the case where minority interests, loan-giving institutions, or joint ventures are of local relevance. Then the headquarters can, on the one hand, enforce the leading GAAP for all management accounting and control purposes on a company-wide basis. Due to the differing local financial accounting regimes, this would then correspond to a separate accounting system design within these subsidiaries. On the other hand, the headquarters can allow for diverse MAS that are based on the respective local GAAP, which would then result in an integrated accounting system design within the subsidiaries. A second topic of international interest is in how far the accruals used for value-based management control purposes are creating a third set of books, which then is established separately from the (financial) accounting model. For example, to calculate the so-called Economic Value Added (EVA) as a residual profit measure under the “economic model of value” (Stewart, 1999, p. 23), it is proposed that research expenses be accounted for as an expenditure. In a survey of 29 U.S. corporations, Weaver (2001) finds on average 19 adjustments to the financial accounting database for EVA measurement, which may be interpreted as indicating a tendency towards only partial integration or even separate MAS design in these firms. Our paper is organized as follows. In Section 2, we review the relevant literature. In Section 3, we develop our research model and the resulting four hypotheses. Section 4 describes the design of our study. In Section 5, information on the measurement of both exogenous and endogenous variables is given. Section 6 presents our results, which have been derived using covariance-based structural equation modelling (SEM). Finally, Section 7 provides a discussion and concludes with some implications for future research.

Our research was motivated by the increasing level of integration of financial and management accounting for management control purposes in German-speaking countries. Until today, relevant research has not clearly shown whether this has a positive impact on controllership effectiveness, and if so, whether the underlying causal inference relates both variables directly, i.e., in a technical fashion, or whether they are related indirectly, i.e., via properties of accounting as a financial language for business communication. In sum, our survey of 149 controller-manager dyads from German top 1500 firms indicates that there is a significant positive impact that is not triggered by a technical mechanism, but fully mediated by the consistency of financial language resulting from an increased level of accounting system integration. Our results add a new flavour to the discussion of whether the integration of financial and management accounting relates to controllership effectiveness. We show that a purely instrumental approach to controllers’ tasks, i.e., just taking the supply-side of producing management accounting information into account, ignores consistency of financial language as a driver for controllership effectiveness from a management perspective. Thus, even though the idea of ‘different costs for different purposes’ under a separate accounting system design has been deemed preferable from an information theory perspective, it does not fully meet management's need for ‘one version of the truth’ to be provided by the overall accounting system. Our findings therefore show that ‘good’ management accounting information is not only characterized by relevance, accuracy, timeliness or technical reliability with respect to a given control problem, but also by consistency from a user-side perspective. When advising management, controllers should therefore take special care not only to establish an understandable link between the accounting information provided for managerial decisions and financial accounting information, but also to emphasize this link in communicating with management. In this context, consistency can also be related to the issue of trust with respect to the use of management accounting information. Even though not explicitly analyzed in our study, consistency could also be interpreted as a mechanism underlying the emergence of trust in management accounting information. For example, Busco et al. (2006), who analyze a case of comprehensive organizational change in the aftermath of a merger, describe the implementation of what we would refer to as an integrated accounting system design as one crucial factor for building trust and therefore facilitating acceptance of new rationales and routines in management. Nevertheless, in professional practice it may be difficult to achieve such consistency, i.e., a consistent financial language under an integrated accounting system, if the relevant financial GAAP system is not appropriate for management control purposes (e.g., if it is mainly driven by tax or legal considerations, as has been the case with German GAAP). Even though our study clearly indicates the relevance of consistency, this property of management accounting information is not a comprehensive substitute, but rather – at least partially – a complement to other properties of ‘good’ accounting information, e.g., information content or relevance for a given decision-making or control problem. As financial accounting standards under IFRS or US-GAAP are more suitable for internal decision-making and/or decision-influencing compared to German, Austrian or Swiss GAAP, an integration of accounting systems is therefore probably easier and more successful. Our results can also be applied to the field of value-based performance measurement. As any adjustment to the financial accounting database reduces consistency, we draw the tentative conclusion that value-based profit measures should not be too sophisticated, but should clearly relate to the underlying profit measures published in financial reports. Nevertheless, this issue should be covered by further research. The notion of consistency is also of importance with respect to the design of management control systems in multinational companies whose foreign subsidiaries act under local financial GAAP regimes. To facilitate management control within these subsidiaries, headquarters could refrain from enforcing the firm's primary GAAP to allow for intra-subsidiary consistency of financial language. This implication should be the subject of further research, as empirical studies provide rather mixed results on whether headquarters indeed follow this practice (e.g., Angelkort et al., 2009 and Hopper et al., 1992) or not (e.g., Joseph et al., 1996). The in-depth analysis regarding the impact of the controllers’ tasks shows that consistency between financial and management accounting information cannot be achieved in a naive fashion by simply using the financial accounting numbers for controllership purposes as well. For example, neither with planning and budgeting nor with performance measurement is there a statistically significant total effect via the mediating variable consistency of financial language. Evidently, there are parts of the management accounting system in which consistency does not seem to matter as strongly as the adaptation of information to control purposes. Consistency, on the other hand, does matter with respect to the firm's internal accounting reports. It is not only established by the controllers’ reports themselves (indicated by the significant total effects regarding the controllers’ tasks ‘reporting’ and ‘accounting information technology design’ as the underlying auxiliary function), but also by a close cooperation between the controllers themselves and the financial accountants. We assume that our results in this respect could be expanded to the implementation of IFRS 8/SFAS 131 on segment reporting. Under the management approach, the segment result has to correspond to the performance measure reported to the segment's chief operating decision-maker. If, for example, a firm uses profit measures based on a separate accounting system design, the resulting transitions to GAAP-based group performance could be perceived as inconsistent by shareholders and other external stakeholders. An interesting topic of further research could thus be how capital markets deal with non-GAAP-based segment performance. Similar to most studies, our findings are subject to limitations. First, it is important to note that our research approach is comprehensive only with respect to the overall controlling function, so that the quality of the accounting information with respect to the specific decision problems at hand has not been established. Hence, our results do not support the recommendation to controllers to relax with respect to information quality per se, but rather to take special care to present the relevant management accounting information in a consistent fashion. Furthermore, our analysis is based on data regarding companies’ top-management level. Therefore, our results have to be interpreted carefully with respect to lower hierarchy levels, even though our theoretical model does not suggest contradictory results in this respect. Nevertheless, as local managers’ information needs typically differ from those of top management, their frame of reference for judging accounting information consistency might be different, calling for other solutions than an integrated accounting system to achieve such consistency. If, for example, non-financial performance measures are used locally, they might be more consistent with a separate management accounting database, using imputed or standardized costs and revenues for performance measurement. In that case, the business unit controller has to shield the local manager from financial accounting information provided to top hierarchy levels using a separate management accounting database for operational purposes (Euske et al., 1993). Such a separate database would not, on the other hand, be reported in a company-wide, aggregated fashion to top management. These considerations might also partly explain the existence of hybrid forms of integrated accounting system design in larger firms. Finally, even though the total effect between the variables integration level of accounting systems and controllership impact on management decisions’ mediated by consistency of financial language is significant, and the path coefficient between the first, exogenous variable and consistency indicates a strong (0.43) as well as highly significant effect, the ratings on ‘integration level of accounting system’ only explain 18% of the variance of consistency of financial language. This leaves ample room for questions on what other effects might drive this variable, which we cannot answer from our study and which would be a matter of future research. Other limitations of our study concern the statistical side. First, due to our non-random sample of firms, findings are limited in terms of representativeness. However, our analysis can be considered representative with regard to the underlying population, which comprises 1269 of the biggest companies in Germany by sales volume. As we draw on cross-sectional data, our findings may not hold for a given type of industry. On the other hand, there are no indications in our theoretical model that the issues discussed in this paper may differ in relevance with respect to specific industries. A second statistical limitation results from the quasi-formative measurement of the variable integration level of accounting systems by means of an index. As this index is measured as a manifest variable, it lacks an error term that regular formative latent variables usually have. This error term represents the impact of all remaining causes other than those represented by the indicators included (Diamantopoulos, 2006). Using the composite thus assumes that the underlying indicators completely capture the construct, which in most cases is inappropriate (Diamantopoulos et al., 2008). However, as Diamantopoulos (2006) points out, this approach is legitimate if all possible indicators of a construct can be conceivably specified. Due to the applied comprehensive measurement drawing on five key controller tasks derived from the relevant literature, this requirement is probably met to a large extent in the case of our composite. Future research should address the design of integrated accounting systems on different hierarchy levels. Additionally, longitudinal studies should be conducted to analyze the consequences of variation in the level of integration of financial and management accounting on controllership effectiveness. In addition, taking into account that this is a crucial issue regarding controllership effectiveness, more effort should be made to identify and analyze other variables that influence the consistency of financial language and/or the consistency of accounting information for specific groups of users.