Performance: the Influence of the Involvement of the Management Accountant

Enterprise Risk Management and Firm

Performance: the Influence of the Involvement of the Management Accountant

By

Esther Rijkelijkhuizen

MSc Business Administration: Organizational & Management Control University of Groningen

Faculty of Economics and Business

18-01-2015

Supervisor: Prof.dr.ir. van Veen – Dirks

Parkweg 149 9727 HB Groningen

The Netherlands

e.m.h.rijkelijkhuizen@student.rug.nl Student number: S2025531

Word count: 11.997

1

Abstract

Over the last decades, the popularity of Enterprise Risk Management (ERM) has risen enormously. ERM is a holistic approach of risk management which is present throughout the entire organization and applied in the organization’s strategy and objectives. ERM is believed to have a positive influence on firm performance, though empirical evidence is scarce. Furthermore, the growing interest in ERM has influenced the management accounting function. This study investigates the relation between ERM implementation and firm performance, and whether this relation is influenced by the extent to which management accountants are involved in ERM. Data is collected with a web survey distributed among risk managers and management accountants. A regression analysis is performed based on 24 observations. The results show a significant positive relation between ERM implementation and firm performance. However, in contrast to the expectations, no evidence is found for a moderating effect of the involvement of the management accountant in ERM.

Keywords: Enterprise Risk Management, Firm Performance, Management Accountant

Introduction

As a result of globalization and other world events, the environment in which organizations operate has become more complex, dynamic and unpredictable (McMullen and Shepherd, 2006).

Therefore, during the past decades, organizations have increasingly adopted risk management practices in order to identify, assess and monitor the risks associated with the increasing complexity and uncertainty (Merchant, 2012). Organizations have become more aware of risk, and increasingly become organized around risk (Soin and Collier, 2013). Hence, the popularity of risk management has risen enormously. The movement towards Enterprise Risk Management (ERM), has placed the emphasis on a more holistic and comprehensive approach of risk management (Soin and Collier, 2013). ERM helps an organization to identify and manage company-wide risks to provide assurance on the achievements of its goals (Gates, Nicolas, and Walker, 2012). The difference with traditional risk management is that it does not treat each risk separately, but it rather manages all risks faced by the organization in a consistent manner (Gordon, Loub, and Tseng, 2009). The popularity of ERM has affected different functions across the organization, in particular the management accounting function. This is illustrated by the fact that the majority of the articles about ERM is published in finance and accounting journals (Bromiley, McShane, Nair, and Rustambekov, 2015). However, despite of the growing interest in ERM and the role of the management accountant in the academic literature, the knowledge regarding this subject is surrounded by inconsistencies and uncertainties (Lundqvist, 2014).

Because of its comprehensive approach of risk management, ERM is expected to improve strategic and operational decision making (Liebenberg and Hoyt, 2003; Nocco and Stulz, 2006).

It helps senior management to oversee the risks faced by the entire organization (Beasley,

Padach, and Warr, 2008), and thereby it improves risk management. Moreover, the efficiency of

the holistic approach can lead to significant cost savings (Grace, Leverty, Philips and Shimpi,

2015; Hoyt and Liebenberg, 2011). Consequently, one would expect a positive relation between

ERM implementation and firm performance. However, empirical studies on the benefits of ERM

implementation show mixed results. The majority of the empirical studies on ERM are based on

2 secondary data. However, risk disclosures relate generally to specific risks and do not provide information on whether risk is managed in a comprehensive manner (Liebenberg and Hoyt, 2003). The use of too simple proxies for ERM implementation might be the cause of the mixed results (Lundqvist, 2014). Therefore, this study further investigates the subject by testing the relation between ERM implementation and firm performance, using survey data instead of secondary data.

Furthermore, this study takes the role of the management accounting function in ERM into account. Nowadays, risk management significantly influences decision making processes. Given the fact that management accounting aims to facilitate in enterprise decision making, management accounting and risk management should complement each other (Rasid, Rahman, and Ismail, 2011). Moreover, the emphasis of risk management has been shifted to management control (Soin and Collier, 2013), which makes it nowadays a domain for management accountants to engage in. Management accountants can contribute by providing information for ERM. Ex-ante information is valuable for risk management decision making while ex-post information can facilitate in risk monitoring (Rasid et al., 2011). Management accountants are also expected to have the skills to evaluate and communicate risk management information better (Williamson, 2004). Therefore, the involvement of management accountants in risk management processes seems to be valuable. However, empirical research regarding this subject is scarce.

Some empirical studies have been conducted on the role of management accounting in ERM, though the results are inconclusive. This study aims to respond to the scarcity of research on this subject on the one hand, and the mixed results on the other. It further examines to what extent and in what ways management accountants are involved in ERM. Moreover, the impact of that involvement on the relation between ERM implementation and firm performance will be investigated.

Consequently, the following research question will be answered: what is the relation between ERM implementation and firm performance, and is this relation influenced by the extent to which management accountants are involved in ERM? This study contributes to the emerging field of research on ERM, by further investigating the relation between ERM and firm performance.

Moreover, it contributes to both the ERM literature and the management accounting literature by investigating the effect of integrating the management accounting function in ERM processes.

Managerial interests are covered by providing information on whether ERM implementation is valuable for organizations, and on whether management accountants should be involved in ERM.

This thesis is structured as follows. First, a review of the literature is provided. Second, the methodology of this study is described, which is divided into data collection procedures, measurement tools, and data analysis. Third, the results are presented. Finally, the results will be discussed followed by a conclusion and limitations of the study.

Theoretical Background and Hypotheses

This section provides a review of the literature, divided into two main parts. First, the concept of ERM is addressed by providing definitions of terms. Furthermore, an overview of the literature on the relation between ERM and firm performance is provided, resulting in the first hypothesis.

Second, the management accounting function is addressed, and the literature on the role of

3 management accountants in ERM is explored. After that, the second hypothesis is developed.

This section results in a conceptual model.

Enterprise Risk Management

Risk is the potential losses or gains, as a result of the uncertain environmental variables associated with taking certain courses of action (Chapman and Cooper, 1983). Over the last years, next to financial risks, the significance of non-financial risks has been recognized as well.

Those risks can be divided into operational risk, reputational risk and strategic risk (Nocco and Stulz, 2006). Operational risk is the potential direct or indirect losses as a result of insufficient internal processes, people or systems (Brown, 2012). Reputational risk is defined as the potential outcomes of shared socially constructed impressions of an organization (Bebbington, Larrinaga, and Moneva, 2008). Strategic risk refers to the consequences of corporate strategic moves (Baird and Thomas, 1985). In this study, the focus is on both financial and non-financial risks.

In order to deal with risk, organizations adopt risk management practices. Risk management is the combination of principles, frameworks and processes that are used for managing risk (Soin and Collier, 2013). It provides a map of risks which forms the base of a plan of action (Millo and MacKenzie, 2009). The goal of risk management is not to minimize the total risk of an organization per se, instead it aims to choose the optimal level of risk to maximize value creation (Meulbroek, 2002). Organizations can manage their risks fundamentally in two ways (Nocco and Stulz, 2006): (1) each risk separately, or (2) all risks together in a consistent manner. The first way is known as traditional risk management, the second way is known as ERM. Traditional risk management is compartmentalized and uncoordinated within an organization (McShane, Nair, and Rustambekov, 2011). It treats each risk individually, as if there is no interaction. It is a fragmented approach, since risk is managed by different responsible business units or departments. Therewith, its focus is on impeding financial risks by using derivatives and insurances (Farrell and Gallagher, 2015; McShane et al., 2011). However, the environment of organizations has become more unpredictable over the last years. It caused an increasing complexity of connected risks within organizations. As a result, there was a call for a holistic approach to manage risk (Farrell and Gallagher, 2015). The answer to that call was ERM.

The movement towards ERM started in the 1990s. Since then, its popularity has risen enormously. The Committee of Sponsoring Organizations of the Treadway Commission (COSO) defines ERM as follows (COSO, 2004):

Enterprise risk management is a process, […] applied in strategy setting and across the enterprise, designed to identify potential events that may affect the entity, and manage risk to be within its risk appetite, to provide reasonable assurance regarding the achievement of entity objectives.

This definition fundamentally captures that ERM is a process that is present throughout the entire

organization, thereby taking an entity-level, portfolio view of risk. In contrast to traditional risk

management, ERM aims to achieve coordinated management of all risks faced by the

organization (McShane et al., 2011). Moreover, ERM goes beyond simply focusing on risk

avoidance. It recognizes the potential value creation of risk exposure, for which the organization

has a strategic advantage over its competitors (Farrell and Gallagher, 2015). ERM is broadly

4 focused. Everyone in the organization views risk management as a part their task, though it is coordinated by senior managers (Rao and Marie, 2007).

COSO developed a framework as well, that aims to provide guidance for ERM implementation.

COSO’s framework is one of the most cited frameworks. Nevertheless, it is also one of the most discussed ones. It is considered to be equivocal, overly theoretical and vague (Beasley, Branson, and Hancock, 2010), which is a problem for individuals involved in ERM implementation. A number of other frameworks involving ERM implementation were developed. However, those frameworks identify different components, which illustrates the overall uncertainty regarding the essential components of ERM (Lundqvist, 2014).

As a response to that uncertainty, Lundqvist (2014) used survey data of 676 firms to investigate how organizations actually implement ERM. Multiple factor analyses resulted in the identification of four components of ERM implementation: (1) general internal environment and objective setting, (2) general control activities and information and communication, (3) holistic organization of risk management, and (4) specific risk identification and risk assessment activities. In short, the ‘general internal environment and objective setting’ component encompasses the presence of a formal strategy, a code of ethics and performance targets for individuals and the organization as a whole. ‘General control activities and information and communication’ refers to the control activities that ensure the use of policies and procedures.

Those first two components form the preconditions for implementing ERM (Lundqvist, 2014).

They are not specific to ERM, but necessary for a well-functioning ERM implementation. The third component, ‘holistic organization of risk management’, refers to formal policies of how risk should be managed, a senior manager with the responsibility to oversee the risk management processes, and frequent and structured updates about risk-related events. This component separates ERM from traditional risk management activities, since it captures the holistic aspect of ERM. Lastly, the fourth component identifies true risk management activities, which are naturally of crucial importance for ERM. Those are related to financial, operational, reputational and strategic risks. The four components are not self-contained. Only if an organization has all four components present and functioning, then ERM is implemented (Lundqvist, 2014). The four components developed by Lundqvist (2014) are based on actual implementation of ERM.

Therefore, those will be leading in this study, and will provide the basis for measuring ERM implementation in organizations.

ERM and firm performance. ERM is considered to be of crucial importance for

corporate governance, since it helps senior management to oversee the risks faced by the entire

enterprise (Beasley et al, 2008). The improved understanding of risk enhances capital efficiency,

resource allocation and equity return (Hoyt and Liebenberg, 2011). Moreover, ERM creates

synergies between several risk management activities, which improve strategic and operational

decision making (Gates et al., 2012; Liebenberg and Hoyt, 2003; Nocco and Stulz, 2006), and

lead to more efficient risk management activities (Sax and Torp, 2013). The goal of risk

management is to find the optimal level of risk exposure. A holistic approach of risk management

aids in achieving this goal, and thereby attempts to create shareholder value (Farell and

Gallagher, 2015). Moreover, the efficiency of the comprehensive approach leads to significant

cost savings (Grace et al., 2015; Hoyt and Liebenberg, 2011). Consequently, one would expect a

5 positive relation between ERM and firm performance. However, empirical studies regarding this relationship show mixed results.

A few studies have been conducted on the value creating capability of ERM for organizations.

Table 1 provides an overview of those studies. Half the studies used Tobin’s Q as the dependent variable, which is a measure of firm performance based on the financial market (Hoyt and Liebenberg, 2011).

Table 1

Previous empirical studies on the value of ERM implementation

Authors ERM identifier Dependent variable Result(s) Beasley et al. (2008) Announcement of

CRO appointment

Shareholder wealth No significant results Farrell and Gallagher

(2015)

Survey responses on ERM maturity

Firm value: Tobin’s Q Significant positive relation

Gates et al. (2012) Survey responses on components of COSO’s framework

Survey responses Enhanced decision- making, management accountability and communication Gordon et al. (2009) Firm disclosure about

ERM activities

Firm performance:

risk and return tradeoff

Depends on

contingency variables Hoyt and Liebenberg

(2011)

Firm disclosure about ERM activities (keyword search)

Firm value: Tobin’s Q ERM significantly enhances firm value McShane et al. (2011) S&P risk management

rating

Firm value: Tobin’s Q No significant results Pagach and Warr

(2010)

Announcement of CRO appointment

Range of financial measures

No significant results Tahir and Razali

(2011)

Firm disclosure about ERM variables

Firm value: Tobin’s Q No significant results

Some studies found evidence for the value creating capability of ERM. For instance, Hoyt and

Liebenberg (2011) performed a keyword search on firm disclosures about ERM activities of 275

insurance companies in the United States. They found that ERM implementation significantly

enhances the firm’s Tobin’s Q by almost 17 percent. A survey by Gates et al. (2012) among 150

risk management executives worldwide, found significant evidence for the value creating

capability of ERM as well. The results indicate that ERM implementation improves decision-

making, management accountability and communication and that it leads to greater management

consensus. Recently, Farrell and Gallagher (2015) published their research on ERM maturity and

the effect on firm’s Tobin’s Q. Based on a survey data of 225 risk management individuals, they

found a significant increase of approximately 25 percent for firms considered as having a mature

ERM approach. Gordon et al. (2009) looked at the subject form a different angle by including

five firm specific factors that are expected to have an influence on the relation between ERM and

performance. They found evidence that the ERM-performance relation is contingent upon the

6 match between ERM and five factors: (1) environmental uncertainty, (2) industry competition, (3) firm complexity, (4) firm size and (5) board of directors’ monitoring.

Other studies did not find significant results on the value of ERM implementation for organizations. For instance, Beasley et al. (2008) and Pagach and Warr (2010) both measured ERM implementation on the basis of the appointment of a Chief Risk Officer (CRO). Beasley et al. (2008) used a sample of 123 financial and non-financial firms, and found no average significant market reaction on announcements of CRO appointment. Tahir and Razali (2011) found a positive relation between ERM and firm’s Tobin’s Q, however their results were not significant. Their study is based on 528 Malaysian publicly listed firms. Pagach and Warr (2010) based their research on 138 CRO announcements, and neither found significant effects on a range of financial measures. McShane et al. (2011), chose for a different ERM identifier: Standard and Poor’s risk management rating. They found an increase in firm value with the implementation of traditional risk management processes. However, firm value did not increase further when implementing ERM. So they did not find support for the value creating capability of ERM.

The conclusion can be drawn that the empirical studies regarding the value creating capability of ERM show mixed results. As illustrated by table 1, the majority of the empirical research is based on publicly available information. However, risk disclosures relate generally to specific risks, and do not necessarily provide information on whether risk is managed holistically (Liebenberg and Hoyt, 2003). The mixed results might be the consequence of the use of too simplistic proxies for ERM implementation (Lundqvist, 2014). Despite of the inconclusiveness of the empirical results, the theoretical underpinnings regarding the benefits of ERM for organizations cannot be ignored.

Therefore, ERM is expected to enhance firm performance. The following hypothesis can be formulated:

Hypothesis 1: a positive relation exists between ERM implementation and firm performance.

The Management Accounting Function

According to the Institute of Management Accounting (IMA), management accounting can be defined as a profession involved in providing expertise in financial reporting and control, partnering management in decision making, and assist management in formulation and implementation of an organization’s strategy (IMA, 2008). Accordingly, management accountants can have multiple tasks within an organization. Those are generally described as scorekeeping, attention-directing and problem-solving (Lambert and Sponem, 2012).

Scorekeeping and attention-directing are concerned with compliance reporting and controlling, while problem-solving focuses on providing information in order to facilitate decision making.

Those two separate roles are typically described as the bean counter role and the business partner role respectively. The bean counter role is oriented on number crunching, maintaining the functioning of the management accounting system and being a watchdog (Järvenpää, 2007). In this case, the management accountant has the responsibility to ensure the correctness of the financial data and the compliance with the organization’s policy (Lambert and Sponem, 2012).

On the other hand, the business partner role is concerned with being actively involved in strategic

and operational decision making (Byrne and Pierce, 2007). It goes beyond simply reporting what

shows up in the income statement and balance sheet. It is about understanding the financial

7 dynamics of the organization (Brands and Holtzblatt, 2015), and using that information to facilitate in decision making.

The role of the management accounting function within an organization depends on several factors. On the one hand, individual characteristics of the management accountant have significant influence. According to Byrne and Pierce (2007), management accountants can affect their own role within an organization depending on their attitudes, personality and personal initiatives. Moreover, personal competences are considerable, since those represent the ability of the management accountant to engage in a certain role (Järvenpää, 2007). However, it is important to view the management accounting function within its environment as well.

Management accounting is shaped by competitive forces and institutions on different levels (Lambert and Sponem, 2012). Moreover, accounting practices are intertwined in the organization’s culture and in a broad range of different practices across different functions (Järvenpää, 2007). So the role of the management accountant is also determined by the expectations and demands of the organization. Nowadays, the management accounting function is more involved in actual business processes and decision making. Over the last decades, software innovations and the development of financial and operational control systems have influenced the management accounting function (Friedman and Lyne, 1997; Byrne and Pierce, 2007; Järvenpää, 2007). As a result, it has become easier to process large amounts of data and routines can be carried out more effectively (Järvenpää, 2007). This is one of the reasons there has been a movement towards the business oriented role of the management accountant, next to for instance decentralization, and innovations in management accounting techniques (Windeck, Weber, and Strauss, 2013).

The role of the management accountant in ERM. The growing interest in ERM has

influenced the management accounting function as well. The two domains of management

accounting and risk management are increasingly and inseparably interdependent (Bhimani,

2009). This is not surprising, given the fact that both domains are integral to performance

measurement and strategic planning (Rasid et al., 2011), and both aim to contribute to enterprise

decision making. Within the literature, several arguments can be found for the potential role of

the management accounting function in ERM. Essentially, management accounting can enhance

ERM in four ways. First, management accountants can contribute by providing both ex-ante

information and ex-post information for risk management purposes. Ex-ante information is

helpful for risk management decision making while ex-post information can facilitate in risk

monitoring (Rasid et al., 2011). Second, management accountants are professionally trained and

experienced in the analysis of information and systems (Collier et al., 2007). Therefore, they are

expected to have better analytic and modeling skills for the evaluation and communication of risk

management data (Collier et al., 2007; Williamson, 2004). In this way, management accountants

can play a significant supporting role in the analysis of ERM information. Third, an important

connection between the management accounting function and the risk management function can

be found in performance measurement systems. According to McWhorter, Matherly and Frizzell

(2006), performance measurement systems enhance ERM by aiding in the identification of risks

and the assessment of risk management information. Moreover, performance measurement

systems can connect risk management to strategic performance measurement (Beasley, Chen,

Nunez and Wright, 2006). The fact that performance measurement is an important part of the

management accounting function illustrates the potential role for management accountants in

8 ERM. Fourth, according to Rasid et al. (2011), possibilities exist for the use of management accounting techniques in risk management since both functions are concerned with costs.

Because of those four arguments, there seems to be an opportunity for management accountants to play a role in ERM.

However, empirical studies regarding the role of the management accounting function in ERM show mixed results. Furthermore, only a few studies have been conducted on this subject. Some studies found evidence for the benefits of integrating the management accounting function in ERM. Collier et al. (2007) conducted four little case studies and a survey among organizations within the United Kingdom. They found that there is marginal integration between the two functions, but found the potential for management accountants to play a more significant role in implementing and developing risk management. The results of Rasid et al. (2011) support that there is a role for management accountants in risk management as well. Their study is based on a mail survey of listed financial institutions in Malaysia, with 72 respondents. They found that the analysis of financial statements contribute most to risk management practices. Mikes (2006) conducted a case study, and found support for the suitability of integrating accounting control and risk control, since they both aid in decision making. On the other side, Soin (2005) found no evidence for the supporting role for management accountants in risk management. Her results suggest that organizations in the financial services sector do not use management accounting techniques for risk management purposes, therefore there is no need for integration between the two functions. Collier et al. (2004) conducted a study in the financial services sector as well.

Their study neither found evidence for the integration of management accounting in risk management. According to Collier et al. (2004), this might be caused by the fact that management accountants are generally viewed as having a more supportive role rather than a managerial role and the stereotypical view that management accountants being more risk averse.

Because of the scarcity of empirical research on the one hand, and the mixed results on the other, conclusions about the role of the management accountant in ERM cannot be drawn. However, the theoretical underpinnings of the potential benefits of involving management accountants in ERM are clearly described in the literature. Consequently, it is expected that the integration of the management accounting function would contribute to the effectiveness of ERM, and would thereby influence the outcomes of ERM implementation for organizations. Hence, one would expect that the involvement of the management accountant will positively influence the relation between ERM and firm performance. Consequently, the following hypothesis can be formulated:

Hypothesis 2: the involvement of the management accountant in ERM has a positive effect on the relation between ERM implementation and firm performance.

Figure 1 shows a conceptual model which graphically illustrates the hypotheses that are proposed.

9

H2

H1

Figure 1. Conceptual model.

Methodology

In this section, the methodology of this study is described. First, the data collection methods and the sample are explained. Second, the measures of the different constructs are addressed. Last, the analysis of data is provided.

Data Collection and Sample

Risk management and the role of the management accountant is to a considerable extent addressed in the academic literature. However, the empirical evidence regarding this subject is still inconclusive. Therefore, theory testing is a suitable approach (Van Aken, Berends and Van der Bij, 2012, p. 14). Theory testing aims to investigate relationships among variables through hypothesis testing based on large-scale data collection (Snow and Thomas, 1994; Van Aken et al., 2012, p. 16). Data is collected using a web survey. The most common form of a web survey is used, which is a computerized questionnaire that is answered without an interviewer being present (De Leeuw and Dillman, 2008, p. 264). The questions are prepared using The International Handbook of Survey Methodology (De Leeuw and Dillman, 2008) in choice of vocabulary, effective responses and question order. The survey is provided in appendix I.

The target population is risk management professionals, management accountants and other professionals involved in risk management. Other studies targeted risk management professionals (Gates et al., 2010; Farrell and Gallagher, 2015) and management accountants (Farrell and Gallagher, 2015) as well. Data is collected using three different methods. First, the survey is spread among the members of nineteen discussion groups on LinkedIn, by placing a request on their pages. The groups are established for network purposes and for discussing subjects relevant to ERM and management accounting. Members of those groups are professionals in relevant disciplines. Eleven groups comprise (enterprise) risk management, with 86.746 members in total.

The remaining eight groups comprise management accounting or controlling, with 402.854 members in total. However, the number of members only indicate how many people could have had access to the survey. It does not indicate how many members actually read request to fill in the survey. Therefore, it is impossible to determine the exact response rate. Second, the link to the survey is spread by e-mail among fifty Dutch qualified controllers, that are members of the Dutch Controlling Institution. Third, the survey is spread among potential respondents within the personal network of the researcher.

ERM implementation

Involvement of the management accountant in ERM

Firm performance

10 The final sample consists of 24 respondents. Four of them had missing data. However, given the fact that both the number of missing items, and the number of respondents with missing items are less than 20 percent, replacing missing data is suitable in this case (Downey and King, 1998).

Missing values are replaced with the mean of all respondents for that item.

Table 2

Functions of respondents

Frequency Percent

Valid Percent

Cumulative Percent

Valid Risk manager 9 37,5 37,5 37,5

Management accountant / controller 7 29,2 29,2 66,7

Consultant 3 12,5 12,5 79,2

Other 3 12,5 12,5 91,7

Internal auditor 2 8,3 8,3 100,0

Total 24 100,0 100,0

Table 2 provides an overview of the functions of the respondents. Nine of them are risk managers, and seven of them are management accountants or controllers. The remaining respondents have another function within the organization they are working. Approximately 46 percent of the respondents is currently working in Europe, 29 percent in Northern America, and the remaining in Asia and Africa. The majority of the respondents is working at a firm with more than thousand employees. The service industry is represented the most (29%), followed by the finance, insurance and real estate industry with almost 17 percent. A complete overview of the characteristics of the respondents can be found in appendix II.

Measurements

The following three concepts need to be elaborated on: ERM implementation, firm performance and the involvement of the management accountant. Because of the fact that a survey will be used in gathering data, only perceptions of the concepts are measured. In addition, control variables are identified. Appendix III shows a complete overview of the constructs and measurement items.

ERM implementation. The four components of ERM, developed by Lundqvist (2014), with the associated identifiers are the starting point for measuring the ERM concept.

Those components are (1) general internal environment and objective setting, (2) general control activities and information and communication, (3) holistic organization of risk management and (4) specific risk identification and risk assessment activities. A selection of the identifiers of the is made, based on other literature measuring the ERM concept as well (Beasley et al., 2010;

Farrell and Gallagher, 2015; Gates et al., 2012; Sax and Torp, 2013; Zhao, Hwang and Low, 2013). Questions related to ERM implementation are answered using a five-point Likert scale.

Statements are made about the presence of certain elements within their organization, and the

respondent is expected to give an indication about the level of agreement or disagreement. It is

expected that the opinion is divided, since most respondents are expected to either have

implemented those elements or have not. Therefore, most answers will be centered around either

disagreement or agreement. In that case, scale reliability is independent of the number of

11 categories (Lavracas, 2008, p. 427), which makes a five-point scale sufficient. Survey questions eight and nine are related to ERM implementation.

Firm performance. In survey question seven, firm performance is measured based on eight items. Four items are general performance measures. Since ERM enhances (1) strategic decision making and (2) operational decision making (Gates et al., 2012; Liebenberg and Hoyt, 2003; Nocco and Stulz, 2006), those items are used for measuring firm performance.

Furthermore, based on Gordon et al. (2009), (3) profitability and (4) exploitation of new opportunities are added as measures. The remaining four performance items are related to risk and risk management. Those items are: (5) efficiency of risk management activities, (6) oversight of risks, (7) understanding of the risks faced by the organization (Beasley et al., 2008), and (8) ability to pick an optimal level of risk exposure (Sax and Torp, 2013). Based on Sax and Torp (2013), the respondents were asked to rate those performance items over the last three years compared to the sector, on a five-point scale. A semantic differential measurement scale is used to identify the perceived performance of the firm compared to the sector (significantly worse versus significantly better). Such a scale measures both the direction and intensity of the respondent’s opinion (Lavracas, 2008, p. 810).

Involvement of the management accountant. Essentially, the management accountant can enhance ERM in four ways: (1) information provision (Rasid et al., 2011), (2) evaluation and communication of risk management information (Collier et al., 2007; Williamson, 2004), (3) integration of risk management in performance measurement systems (Beasley et al., 2006;

McWhorter et al., 2006), and (4) the use of management accounting techniques in risk management (Rasid et al., 2011). Those four ways form the base for six measurement items in survey question ten. The involvement of the management accountant is measured on a five-point Likert-type scale, representing how often the management accountant is involved in risk management activities. Respondents are asked how often those potential ways of involvement take place on a five-point scale (never, seldom, sometimes, often or always).

Control variables. According to Gordon et al. (2009), firm performance and its relation

to ERM is contingent upon contextual factors, including: (1) firm size, (2) firm complexity and

(3) environmental uncertainty. Therefore, those variables are used as control variables and are

processed in survey questions. Respondents are asked about the size of the organization they are

working at. This factor is measured in categories, so on a nominal scale. Consequently, dummy

variables are created. Final scores are formed on small sized (less than 100 employees), medium

sized (between 100 and 500 employees) and large sized (more than 500 employees) firms. Firm

complexity is measured by the number of business segments of the firm (Gordon et al., 2009),

which is measured on a nominal scale as well. Final scores are formed on low complexity (one or

two segments), medium complexity (three or four segments) and high complexity (five or more

than five segments). For environmental uncertainty, the measurement scale of Govindarajan

(1984) is used. Respondents are asked about the predictability of market demand, government

regulations, labor union actions, competitors’ actions and technology developments, measured on

a five-point scale. Survey questions three, five and six represent the control variables.

12 Data Analysis

In order to group the questions from the questionnaire and to check if those questions empirically reflect the constructs that are intended to be measured, a confirmatory factor analysis is conducted. Since the goal is to convert the original set of variables into a set of uncorrelated variables (Dunteman, 1989, p. 7), and to reduce the dimensionality of the data (Richardson, 2009), a principal component analysis is conducted. The purpose is to identify factors of which the components have loadings higher than 0.40 on only one factor.

First, the analysis of the 23 ERM-items is conducted. The goal is to find a four factor solution, since the four components of ERM of Lundqvist (2014) form the base of measuring ERM. Table 3 provides a description of the final items in the four factor solution. The Kaiser-Meyer-Olkin measure of sampling adequacy is 0.820, which is higher than the required minimum of 0.5 (Leech, Barrett, and Morgan, 2005, p. 94). Moreover, the Bartlett’s test of sphericity is significant (p < 0.05), which represents the significance of all the correlations within the correlation matrix (Leech et al., 2005, p. 94). The four separate factors explain respectively 11.4, 5.0, 61.4 and 8.3 percent of the total variance. Twelve items had to be deleted since they contained cross-loadings higher than 0.40.

Table 3

Principal component analysis with Oblimin rotation for 23 ERM related survey items

1 2 3 4

Item

Internal environment

General control

Holistic organization

Risk management

Formal strategy ,775

Performance targets ,631

Formal responsibilities ,960

Authorization processes ,427

Formal communication channels ,615

Statement of risk appetite ,989

Determined quantitative impacts ,888

Chief Risk Officer ,777

Board’s subcommittee ,768

Communication of importance of risk ,871

Reputation risk exposure 961

Eigenvalues 1.517 0.715 11.546 1.032

Percentage of total variance 11.401 4.978 61.379 8.301

Cronbach’s Alpha .817 .789 .945

Factor 1 is related to the first component of Lundqvist (2014), which describes the general

internal environment and objective setting. One out of four survey items had to be deleted

because of high cross-loadings. The final factor 1 describes respectively the existence of a formal

strategy, performance targets, and formally defined responsibilities for the executive

management. Factor 2 relates to general control activities and information and communication,

which is the second component of Lundqvist (2014). This factor consists of authorization

processes to ensure the use of policies, and procedures and formal communication channels to

report suspected breaches of law and regulations. Two out of four survey items related to this

13 component had to be deleted as well. The third factor relates to the third component of Lundqvist (2014), which is the holistic organization of risk management. Because of high cross-loadings, only five out of eleven items can be included in this factor. Those are the existence of a statement of risk appetite, determined quantitative impacts of risk, a Chief Risk Officer, a board’s subcommittee, and the communication of the importance of risk among employees. The fourth factor, related to the fourth component of Lunqvist (2014), comprises of the identification of reputation risk exposures. The other three survey items related to risk identification and risk assessment activities had to be deleted. A reliability analysis on the first three factors is performed, by measuring the Cronbach’s Alpha which represents the internal consistency reliability (Leech et al., 2005, p. 87). The Cronbach’s Alpha’s are respectively .817, .798 and .945, which are all above the required minimum of 0.70. It is not possible to determine the Cronbach’s Alpha of factor 4, since that factor contains of only one item.

Second, the analysis of six items related to management accountant involvement is conducted, which resulted in a solution with one factor explaining 69,6 percent of the total variance. The Keiser-Meyer-Olkin measure of sampling adequacy was 0.825, which is higher than the required minimum. The Bartlett’s test of sphericity was significant (p < 0.05). The loadings of the items are higher than 0.40 so no item had to be deleted. Consequently, the final factor comprises the provision of respectively ex-ante and ex-post information, the evaluation and communication of risk management information, and the use of performance measurement systems and management accounting techniques in risk management. The Cronbach’s Alpha is 0.912, which is higher than the required 0.70. Table 4 provides an overview of the final factor and its items.

Table 4

Principal component analysis for six management accounting related survey items 1

Item

Management accountant involvement

Ex-ante information .845

Ex-post information .876

Evaluation of risk management information .831 Communication of risk management information .847

Performance measurement systems .793

Management accounting techniques .812

Eigenvalue 6.803

Percentage of total variance 69.557

Cronbach’s Alpha .912

Last, final construct scores on both ERM implementation and management accountant involvement are formed, by calculating mean scores of the final factors identified in the principal component analysis. Moreover, separate constructs are formed for the four components of ERM implementation, by calculating averages as well. In order to be able to perform a moderator analysis, an additional construct is formed that accounts for the potential interaction effect of ERM implementation and management accountant involvement. The construct comprises the multiplication of the scores on ERM implementation and management accountant involvement.

In moderation analyses, mean centering the predictor variables is advantageous since it reduces

14 problems of collinearity and it improves the interpretability of the estimates (Little, Card, Boivaird, Preacher, and Crandall, 2007). Therefore, mean-centered scores are used in testing the second hypothesis.

Results

This section provides an overview of the empirical results. The descriptive statistics and correlation table are shown at first. After that, the regression analyses for each hypothesis are shown separately. Last, the results of two additional analyses are presented.

Descriptive Statistics and Correlations

Table 5 provides an overview of the descriptive statistics, divided into three parts. The first part provides descriptive statistics on the main variables of this study. The second part contains of the four components of ERM implementation. The last part provides information on the six items measuring the involvement of the management accountant in ERM.

Table 5

Descriptive Statistics

N Minimum Maximum Mean

Std.

Deviation

Percentiles

25 50 75

Firm_performance 24 2.38 4.88 3.58 0.672 3.06 3.50 3.97 ERM_implementation 24 2.06 5.00 3.61 0.852 2.77 3.83 4.07

MA_involvement 24 1.00 4.50 2.84 1.064 2.04 2.75 3.79

ERMxMA 24 3.64 21.45 10.66 5.616 5.63 9.63 14.79

Internal_environment 24 2.33 5.00 4.11 0.807 3.67 4.33 4.67

General_control 24 1.00 5.00 3.83 1.013 3.50 4.00 4.46

Holistic_organization 24 1.20 5.00 3.16 1.317 1.90 3.40 4.46

Risk_management 24 1.00 5.00 3.33 1.167 3.00 3.00 4.00

Exante_information 24 1.00 5.00 2.75 1.359 2.00 2.00 4.00 Expost_information 24 1.00 4.00 2.75 1.225 2.00 3.00 4.00

Evaluation 24 1.00 5.00 2.71 1.233 2.00 3.00 4.00

Communication 24 1.00 5.00 2.70 1.300 1.25 2.90 4.00

Performance_measurement 24 1.00 5.00 3.18 1.307 2.00 3.60 4.00 Accounting_techniques 24 1.00 5.00 2.96 1.233 2.00 3.00 4.00 Valid N (listwise) 24

The average firm performance is rated 3.58 on a five-point scale. The average score on ERM implementation is quite similar, 3.61, also measured on a five-point scale. On the other hand, the mean of the involvement of the management accountant is lower and has a higher standard deviation. On average, firms have the highest scores on the first and second components of ERM implementation. However, standard deviations are higher for the third and fourth component. The involvement of the management accountant is measured on a five-point scale, representing frequencies. The first four items have a mean around 2.70, which is between “seldom” and

“sometimes”. The mean of the last two items is slightly higher, around 3.00. Furthermore, the

data of the main variables is not normally distributed. In contrast to the expectations, the data is

neither centered around agreement or disagreement.

15 Table 6

Pearson correlation coefficients

Firm_ ERM_ Internal_ General_ Holistic_ Risk_ MA_

performance implementati on

environ ment

control organization manageme nt

involve ment Firm_

performance 1 ERM_

implementation

,683 ^** 1

Internal_

environment

,359 ,744 ^** 1

General_

control

,435 ^* ,854 ^** ,713 ^** 1

Holistic_

organization

,731 ^** ,898 ^** ,577 ^** ,745 ^** 1 Risk_

management

,545 ^** ,651 ^** ,211 ,290 ,446 ^* 1

MA_

involvement

,478 ^* ,479 ^* ,172 ,283 ,566 ^** ,395 1

* Correlation is significant at the 0.05 level

** Correlation is significant at the 0.01 level

The values of Pearson correlations coefficients are reported in table 6. All correlations have the expected signs. ERM implementation, three ERM components and involvement of the management accountant are all significantly correlated with firm performance. Moreover, ERM implementation is significantly correlated with the involvement of the management accountant in ERM. Furthermore, the holistic organization of risk management is significantly correlated with the other three components of ERM implementation.

Hypothesis 1

In order to test the first hypothesis, a linear regression is calculated to predict firm performance based on ERM implementation. Firm size, firm complexity and environmental uncertainty are included as control variables. In the first model, only the control variables are included as predictors for firm performance. In the second model, ERM is added as a predictor. The results of this analysis are provided in table 7.

The R-squared increases from .428 to .704 in the second model compared to the first. The F- change is significant, which means that the variable added in model two significantly improved the prediction. The first model is not significant (p = 0.055). In contrast, on the second model a significant regression equitation was found (F(6,17) = 6.736, p < 0.01).. ERM implementation significantly predicts firm performance in the second model (Beta = 0.625, t(23) = 3.983, p <

0.05). Given the fact that model two accounted for significantly more variance in firm

performance than model one, it can be said that ERM implementation is positively related with

firm performance. Consequently, hypothesis 1 is supported.

16 Table 7

Regression analysis on ERM implementation and firm performance

Model 1 Model 2

Variables Beta t Sig. Beta t Sig.

Intercept 4.082 .001 ^** 2.050 .056

Firmsize ^a .326 1.763 .095 .306 2.237 .039 ^*

Firmsize_medium .004 .019 .985 .263 1.657 .116

Firmcomplexity ^b -.271 -1.040 .312 -.255 -1.324 .203

Firmcomplexity_medium -.366 -1.655 .115 -.188 -1.107 .284 Environmental_uncertainty .384 1.855 .080 .216 1.360 .192

ERM_implementation .625 3.983 .001 ^**

R-squared .428 .704

Adjusted R-squared .269 .599

R-squared change .428 .276

F for change in R-squared 2.689 15.866 ^**

F 2.689 6.736

Sig. .055 .001 ^**

a. 0 = small, 1 = large b. 0 = low, 1 = high

* Significant at the 0.05 level

** Significant at the 0.01 level Hypothesis 2

In order to test the second hypothesis, a linear regression is calculated to predict firm performance based on ERM implementation and involvement of the management accountant. In the first regression model, firm size, firm complexity, and environmental uncertainty are included as control variables. The second model includes the mean-centered scores on ERM implementation and management accountant involvement separately. The third model includes the interaction of those mean-centered variables.

As shown in table 8, the second model shows a significant regression equitation (F(7,16) = 5.437, p < 0.05). The R-squared significantly increases with .276. On the third model, a significant regression is found as well (F(8,15) = 5.020, p < 0.05). The R-squared increases with .024.

However, this increase is not significant (p = 0.268). ERM implementation significantly predicts

firm performance in model two (Beta = 0.615, t(23) = 2.938, p < 0.01), and three (Beta = 0.661,

t(23) = 3.132, p < 0.01). The involvement of the management accountant is not significant in both

models, and shows Betas of .017 and -0.44 respectively in model two and three. The interaction

variable of ERM implementation and involvement of the management accountant neither

significantly predict firm performance (Beta = 0.202, t(23) = 1.151, p = 0.268). It indicates that

there is no significant moderation between ERM implementation and involvement of the

management accountant. Consequently, hypothesis 2 is rejected.

17 Table 8

Regression analysis on ERM implementation, involvement of the management accountant and firm performance

Model 1 Model 2 Model 3

Variables Beta t Sig. Beta t Sig. Beta t Sig.

Intercept 4.082 .001 ^** 5.896 .000 ^** 5.561 .000 ^**

Firmsize ^a .326 1.763 .095 .304 2.141 .048 ^* .191 1.113 .283 Firmsize_medium .004 .019 .985 .256 1.369 .190 .252 1.360 .194 Firmcomplexity ^b -.271 -1.040 .312 -.246 -1.064 .303 -.182 -.773 .452 Firmcomplexity_

medium -.366 -1.655 .115 -.191 -1.063 .303 -.140 -.763 .457 Environmental_

uncertainty .384 1.855 .080 .216 1.317 .206 .235 1.439 .171 ERM_

implementation .615 2.938 .010 ^** .661 3.132 .007 ^**

MA_involvement .017 .077 .940 -.044 -.192 .850

ERMxMA .202 1.151 .268

R-squared .428 .704 .728

Adjusted R-squared .269 .575 .583

R-squared change .428 .276 .024

F for change in

R-squared 2.689 7.472 ^** 1.326

F 2.689 5.437 5.020

Sig. .055 .002 ^** .004 ^**

a. 0 = small, 1 = large b. 0 = low, 1 = high

* Significant at the 0.05 level

** Significant at the 0.01 level Additional Analyses

In order to get some extra information out of the data, two additional analyses are performed.

First, an additional regression analysis is performed on the relation between the four components of ERM separately, and firm performance. Second, two additional regression analyses are performed on ERM implementation and firm performance. In those analyses, general performance measures and risk-related performance measures are separated.

Table 9 provides an overview of the results of the regression with the four components of ERM separately. The second model shows a significant regression equitation (F(9,14) = 6.946, p <

0.01). The R-squared increases significantly with .333 (p < 0.05). Looking at the components of

ERM, only ‘holistic organization’ significantly predicts firm performance in model two (Beta =

0.657, t(23) = 2.689, p < 0.05).

18 Table 9

Regression analysis on components of ERM implementation and firm performance

Model 1 Model 2

Variables Beta t Sig. Beta t Sig.

Intercept 4.082 .001 ^** 2.575 .022 ^*

Firmsize ^a .326 1.763 .095 .283 1.586 .135

Firmsize_medium .004 .019 .985 .197 .952 .357

Firmcomplexity ^b -.271 -1.040 .312 -.350 -1.728 .106

Firmcomplexity_medium -.366 -1.655 .115 -.293 -1.570 .139 Environmental_uncertainty .384 1.855 .080 .055 .299 .769

Internal_environment -.034 -.160 .875

General_control -.049 -.157 .877

Holistic_organization .657 2.689 .018 ^*

Risk_management .121 .668 .515

R-squared .428 .761

Adjusted R-squared .269 .607

R-squared change .428 .333

F for change in R-squared 2.689 4.874 ^*

F 2.689 6.946

Sig. .055 .004 ^**

a. 0 = small, 1 = large b. 0 = low, 1 = high

* Significant at the 0.05 level

** Significant at the 0.01 level

Appendix IV shows the results of the regression analysis in which firm performance is divided into risk-related performance and general performance. In the regression on ERM implementation and general performance, the R-squared increases significantly with .218 (p <

0.01) in the second model. The regression equitation is significant as well (F(6,17) = 4.481, p <

0.01). Moreover, ERM implementation significantly predicts general performance in model two (Beta = 0.556, t(23) = 3.095, p < 0.01). The regression on ERM implementation and risk-related performance does not show a significant regression equitation (F(6,17) = 1.858, p = 0.147).

However, the R-squared increases significantly with .169 (p < 0.05), and ERM implementation significantly predicts risk-related performance (Beta = 0.488, t(23) = 2.179, p < 0.05).

Discussion and Conclusion

This study investigates 24 observations on ERM implementation, involvement of the management accountant in ERM, and firm performance. Using a regression analysis, this study examines the relation between ERM implementation and firm performance, and to what extent this relation is influenced by the involvement of the management accountant in ERM.

ERM implementation and Firm Performance

The results show a significant positive relation between ERM implementation and firm

performance. This is in line with the first hypothesis. Several explanations regarding this relation

are provided in the literature, concerning different facets that directly or indirectly influence the

performance of a firm. First, the holistic approach of risk management induces an improved

understanding of risks faced by the entire enterprise (Beasley et al., 2008). Moreover, it enhances

19 operational and strategic decision making (Gates et al., 2012; Liebenberg and Hoyt, 2003; Nocco and Stulz, 2006). ERM contributes in finding the optimal level of risk exposure, which can create shareholder value (Farell and Gallagher, 2015). Due to the synergies created between different risk management activities, risk management activities are more efficient (Sax and Torp, 2013) which can lead to significant cost savings (Grace et al., 2015; Hoyt and Liebenberg, 2011). Last, according to Hoyt and Liebenberg (2011), ERM enhances capital efficiency, equity return and resource allocation.

Only three studies found a significant positive relation between ERM implementation and firm performance as well (Farrell & Gallagher, 2015; Gates et al., 2012; Hoyt and Liebenberg, 2011).

Two out of those three studies are also based on survey responses (Farrell & Gallagher, 2015;

Gates et al., 2012). Others did not find evidence for the relation between ERM implementation and firm performance (Beasley et al., 2008; McShane et al., 2011; Pagach and Warr, 2010; Tahir and Razali, 2011). Those studies are all based on publicly available information. However, according to Liebenberg and Hoyt (2003) and Lundqvist (2014), risk disclosures do not provide information on whether risk is managed holistically. Therefore, that might be too simple proxies of ERM implementation. It is at least remarkable that there is a clear separation in methodology between the group of studies that found significant results, and the group that did not. It indicates that the methodology of the study has an influence on the outcomes, and might even confirms that secondary data is not suitable for measuring ERM implementation. However, considering the low number of studies, one should be careful drawing conclusions. On the other side, the study of Gordon et al. (2009) has a different outcome. They found that the relation between ERM implementation and firm performance depends on five factors. This study controlled for three out of those factors, namely (1) environmental uncertainty, (2) firm size, and (3) firm complexity.

Only firm size has a significant influence on the relation between ERM implementation and firm performance. It confirms the finding of Gordon et al. (2009) that the ERM-performance relation is dependent on the match between ERM implementation and firm size.

When looking at the four components of ERM separately, it can be concluded that on average there are higher scores on the first two components compared to the last two. The first component represents strategy setting and performance targets, while the second component represents control activities (Lundqvist (2014). Both components together form the preconditions of ERM implementation. Apparently, the preconditions are more often present within organizations. The third and fourth component encompass respectively the holistic management of risk, and specific risk management activities (Lundqvist, 2014). The spread of the third component is higher than the other components, which means that there are larger differences in scores on that component.

Moreover, the holistic management of risk correlates highly and significantly with the other three components. The correlation with strategy setting and performance targets seems obvious, since ERM should be applied in strategy setting, and should be concerned with the achievement of the organizational goals (COSO, 2004; Farrell and Gallagher, 2015). ERM aims to achieve coordinated management of risks (McShane et al., 2011). Therefore, control activities are important, which are represented in the second component. Lastly, the correlation with risk management activities is clear, given the fact that ERM is a holistic manner of managing risk.

Furthermore, the additional regression on the four components of ERM and firm performance

provides extra insights. Only the third component, which is the holistic management of risk,

significantly predicts firm performance. This component separates ERM from traditional risk

20 management activities, since it captures the coordinated management of risks (Lundqvist, 2014).

That holistic aspect has a positive relation with firm performance, which represents the value of ERM over traditional risk management.

It is important to notice the considerably high R-squared in the regression of ERM implementation and firm performance. It means that a high percentage of variance is explained by the model. A potential explanation is that firm performance is partly measured using risk- related items. Therefore, it might be more likely that a higher percentage of variance in firm performance is explained by ERM implementation. However, an additional analysis is performed, that rejects this statement. The analysis shows that ERM implementation significantly correlates with both risk-related performance and general performance. Furthermore, it is important to be careful with assuming causality between ERM implementation and firm performance. A regression analysis only tests the strength and the direction of relationships between variables, which does not necessarily imply causality. For instance, high performance firms might have more resources to successfully implement ERM. In that case, the ERM- performance relationship is the other way around, and also depends on a third variable.

Consequently, the results do not with certainty show that ERM implementation will cause a higher firm performance. They only show a significant positive relation.

Involvement of the Management Accountant in ERM

According to the literature, there are four ways in which the management accountant can be involved in ERM, namely (1) provision of ex-ante and ex-post information (Rasid et al., 2011), (2) evaluation and communication of risk management data (Collier et al., 2007; Williamson, 2004), (3) the use of risk management information in performance measurement systems (Beasley et al., 2006; McWorther et al., 2006), and (4) the use of management accounting techniques in risk management (Rasid et al., 2011). The average scores and the spread of the data indicate that the management accountant is barely involved in both the provision of information, and the evaluation and communication of risk management data. The answers on the frequency of that ways of involvement, are on average between “seldom” and “sometimes”. The average scores on the third and fourth way of involvement are higher, around “sometimes”. Collier et al.

(2007) and Soin (2005) also found that the management accounting function and ERM are barely integrated with each other. Moreover, the standard deviations for all four ways of involvement are high, which means that there are large differences in the involvement of the management accountant. A potential explanation for those differences is that the role of the management accountant in the organization is shaped by personal characteristics (Byrne and Pierce, 2007;

Järvenpää, 2007), and by expectations and demands of the organization (Järvenpää, 2007;

Lambert and Sponem, 2012). The role of the management accountant in ERM might be shaped by the personal and organizational characteristics as well. Perhaps in the future, the role of the management accountant in ERM becomes more important. Due to the increasing complexity of the organization’s environment, there is a call for holistically managing risk (Farrell and Gallagher, 2015). The organization’s increasing demand for ERM implementation might influence the role of the management accountant.

The results show that the involvement of the management accountant in ERM does not significantly moderate the relation between ERM implementation and firm performance.

Consequently, the second hypothesis is rejected. Soin (2005) and Collier et al. (2004) neither