To Disclose or Not to Disclose? An Investigation of the Antecedents and Effects of Open Book Accounting

Abstract

Open book accounting (OBA) is the regular disclosure of management accounting information beyond corporate borders. Prior contributions have mainly concentrated on identifying its antecedents in individual or small numbers of organizations with exploratory cases. My paper responds to the call to investigate OBA on a wider empirical basis and focuses simultaneously on the explanatory variables of OBA and its influence on both financial and non-financial performance. I thus also explore the mediating role of OBA in linking key antecedents and performance within a unified theoretical framework. I empirically test my model using survey data from a sample of European companies, which are then analyzed through structural equation modeling. My findings indicate that the extent of OBA use is explained by a firm’s willingness to work together with its counterparts in the long run, that is, a relational factor, and the presence of sophisticated cost accounting systems, that is, a technical prerequisite. My evidence also suggests a positive association between OBA and firm performance. Additionally, I find that OBA is a partial mediator that explains how a firm’s long-term commitment to its external partners and the sophistication of its cost accounting system may become associated with performance.

ORCID

Ariela Caglio http://orcid.org/0000-0002-5829-9784

Notes

1 To cover multiple relationships at the same time, researchers can leverage on social network analysis techniques, which would require focusing on a single network of entities (Caglio & Ditillo, 2012). However, my research questions entail the analysis of a cross section of firms’ OBA practices.

2 I am grateful to the Associate Editor for suggesting this point.

3 I would like to thank an anonymous reviewer for suggesting the above clarification.

4 In focusing on the performance of the disclosing company, I complement prior literature that specifically concentrates on total cost of ownership. This stream of studies, mainly experiment based, suggests that communicating cost information might improve performance. These contributions are more concerned with the enhancement of mutual performance (e.g. Degraeve, Labro, & Roodhooft, 2005; Masschelein, Cardinaels, & van den Abbeele, 2012; Wouters et al., 2005) and show that several aspects are crucial in this respect, including the power differential and the buyer’s dependence on the supplier (Van den Abbeele et al., 2009), the type of accountability of the counterparts during negotiation (Chang et al., 2013), market pressures, willingness to cooperate, and the perceived risk of opportunistic behavior (Drake & Haka, 2008).

5 To better understand the extent of OBA use and how I measure it, it is important to anticipate here that I employed partial least squares (PLS) analysis to formally examine its construct validity. In doing so, I observed that the different possible pieces of management accounting information that a firm can transfer through OBA are highly correlated (see the item loadings in Table 3). This seems to suggest that it is the decision to implement OBA that leads to the disclosure of those pieces of information reported in Table 3; that is, if a company decides to go open book, it may disclose its revenue, raw material costs, labor costs, and so forth. Hence, the OBA construct is reflective.

6 The SCOR model is a cross-industry framework for evaluating and improving supply chain performance and management. SCOR includes three major segments: process modelling, performance measurement, and supply chain best practices. It provides standard process definitions, terminology, and metrics and enables companies to benchmark themselves against world-class companies (SCC, ‘Supply-chain operations reference-model’, Overview of SCOR version 5.0, 2008).

7 The question was framed as follows: ‘Please check the following statements [1 = strongly disagree, 7 = strongly agree]: (a) The business relationships with our most important counterparts are successful, (b) the business relationships with our most important counterparts contribute to improve our company’s profitability, and (c) the business relationships with our company contribute to raise the profitability of our business partners.’ These items are not included in Table 3.

8 The SCC European members are the European associates of the SCC, a non-profit organization whose aim is to advance the state of the art in supply chain management systems and practices. The AIDA database is provided by Bureau van Dijk and includes companies’ financial statement data and ratios and other details (e.g. the company address and website).

9 In doing so, I referred to the threshold defined by the European Union for the definition of small and medium-sized enterprises.

10 To treat the remaining missing data, comprising less than 5% after the sample is cleaned up, I used the mean replacement function in Smart PLS, as suggested by Hair et al. (2014).

11 The principal component analysis with varimax rotation revealed the presence of six factors with an eigenvalue greater than 1.0. The six factors together account for 73.4% of the total variance; the first (largest) factor does not account for a majority of the variance (28.4%).

12 The two first-order components, financial performance and non-financial performance, form the second-order component Performance. Such higher order modelling involves summarizing the first-order components, or LOCs, into a single multidimensional second-order component, or HOC (e.g. Ringle, Sarstedt, & Straub, 2012). I choose this modelling approach to make the PLS path model more parsimonious and easier to grasp, as suggested by Hair et al. (2014).

13 As a robustness check for my measure, I also correlated Performance with the firms’ earnings before interests and taxes (EBIT; coeff. = 0.48, p = .002) and returns on capital employed (ROCE; coeff. = 0.31, p = .064).

14 In the first stage, the repeated indicator approach is employed to obtain the latent variable scores for the LOCs, which, in the second stage, are used as manifest variables in the HOC measurement. This approach allows other latent variables as predecessors to explain some of the HOC variance, which may result in significant path relationships (Hair et al., 2014). See also Henseler and Chin (2010) and Ringle et al. (2012) on this point.

15 Although distinct, the two relational factors are highly correlated. To make sure that this high correlation does not cause one factor (specifically, propensity to collaborate) to be insignificant, the analysis was rerun with one relational factor at a time. The results confirm that propensity to collaborate is insignificant and that the associations between OBA and performance with long-term orientation are positive and significant.

16 In case of partial mediation, it is possible to specify the proportion of the variance of a dependent variable explained by an independent variable indirectly, via a mediator variable, by calculating the VAF. The VAF indicates the magnitude of the indirect effect in relation to the total effect. It is calculated by dividing the indirect effect coefficient by the total effect coefficient, which is the sum of the indirect and direct coefficients (Preacher & Hayes, 2008).

17 To control for size, I also use the log transformation of the number of employees, total assets, and revenue. The main (untabulated) results remain unchanged except for the association between long-term orientation and OBA, that becomes slightly insignificant (p = .139) when I use the logarithm of total assets. This might suggest a difference in the explaining variables of OBA and performance of capital-intensive businesses.

18 The variable is equal to one if information flows include items OBA2 through OBA17, reported in Table 3. Based on this classification, the number of firms disclosing cost information is 100.

19 PLS MGA refers to a set of techniques for comparing PLS model estimates across groups of data, usually to explore differences between path coefficients in the structural model (Hair et al., 2014, p. 247). PLS MGA is a new functionality embedded in Smart PLS 3.0, together with nonparametric procedures to execute such analyses on data that are not normally distributed.

20 It is important to recognize that, in performing such analysis, I could not rule out the potential confounding effect of a different quantity of information disclosed by firms.

21 When we address interorganizational settings, performance becomes an issue. Firm-level performance is relatively more ‘fact based’ but does not take into account whether OBA is useful in improving the performance of the partners. The relationship performance used in these additional analyses is more aligned with the so-called interorganizational nature of OBA practices, but is far more difficult to measure where problems are also related to the point of view adopted to collect such a measurement. Aware of the advantages and disadvantages of both approaches, I present both, emphasizing the consistency of the results obtained.