The relationship between recognised intangible assets and voluntary intellectual capital disclosure

Transcription

1 The relationship between recognised intangible assets and voluntary intellectual capital disclosure Frank Schiemann (corresponding author) University Hamburg Von-Melle-Park Hamburg Germany Phone Fax frank.schiemann@wiso.uni-hamburg.de Kai Richter Technische Universität Dresden Thomas Guenther Technische Universität Dresden lehrstuhl.controlling@tu-dresden.de This is a non-final version of the article accepted for publication in Journal of Applied Accounting Research (2015), Vol. 16, Issue 2, pp (doi: Please only cite the final article (see

2 The relationship between recognised intangible assets and voluntary Abstract intellectual capital disclosure Purpose: The capitalisation of intangible investments is discussed controversially in the financial accounting literature. International accounting standards are concerned with this issue and generally demand more intellectual capital to be recognised on the face of the balance sheet. If investors and analysts already gather monetary information about intangible assets from the financial report and find such information useful, then the necessity to complement such information with voluntary intellectual capital disclosure will diminish. Accordingly, there should be an association between recognised intangible assets and voluntary intellectual capital disclosure. The paper aims to discuss these issues. Design/methodology/approach: The authors analyse the voluntary disclosure of 264 investor conference and roadshow presentations of German DAX 30 firms in the year 2001, 2003, 2005, and The authors apply regression models to analyse the association between recognition of intangible assets and voluntary intellectual capital disclosure and control for other determinants of voluntary disclosure. Findings The authors find that the magnitude of recognised intangible assets is significantly and negatively associated with the quantity and quality of voluntary intellectual capital disclosure. The authors show that this association is mainly driven by goodwill accounting. In more detailed analyses we find different directions (positive, negative and insignificant) of this relationship for different categories of intellectual capital. Research limitations/implications: Future studies on voluntary intellectual capital disclosure need to consider recognised intangible assets as a determinant to avoid omitted variable problems. Practical implications: The authors provide descriptive evidence about voluntary intellectual capital disclosure practice of Germany s largest firms. Originality/value: The paper provides primary evidence on the association between recognised intangible assets and voluntary intellectual capital disclosure. Keywords Voluntary reporting, Intangible assets, Goodwill, Intellectual capital reporting, Private channel communication 1

3 1. Introduction The accounting literature often argues that voluntary disclosure of financial and nonfinancial information on intellectual capital investments substitutes for the scarce recognition of intangible assets in financial reporting (Basu and Waymire, 2008, Boesso and Kumar, 2007, Botosan, 1997, Gelb, 2002). Following this statement, there should be an association between mandatory disclosure of intangible assets and voluntary intellectual capital disclosure, which intuitively means that voluntary disclosure substitutes for mandatory disclosure. While analytical papers support the notion that an association exists, they also show that the direction of the relationship between mandatory and voluntary disclosure is ambiguous (Bagnoli and Watts, 2007, Einhorn, 2005, Dye, 1990). Remarkably, empirical research on the determinants of voluntary intellectual capital disclosure does not acknowledge any association with the mandatory disclosure of intangible assets, which may create an omitted variable problem. We address this issue and specifically analyse whether and how voluntary intellectual capital disclosure is connected to the magnitude of intangible assets recognised on the balance sheet. We examine a set of roadshow and investor conference presentations to analyse firms voluntary disclosure (García-Meca et al., 2005). In this setting, we can investigate how and to what extent firms provide intellectual capital information to the specific target groups of analysts and investors, which are addressed by such presentations. For the purpose of this paper, we broadly define intellectual capital as the entity of a firm s nonphysical claims to future benefits (Lev, 2001). Intangible assets is a subset of intellectual capital and is narrowly defined as those capitalised values shown on the balance sheet that are related to intellectual capital. [1] 2

4 Specifically, we analyse whether and how the portion of intangible assets, including and excluding goodwill, is related to the quantity and quality of voluntary disclosure in roadshow and investor conference presentations. The evaluation of voluntary disclosure is based on 264 roadshow and investor conference presentations of German firms of the equity index DAX 30 in the years 2001, 2003, 2005 and We calculate the quantity of voluntary disclosure based on Garcia-Méca et al. (2005) and the quality based on Boesso and Kumar (2007). We believe that roadshow and investor conference presentations provide an optimal setting for our analysis because firms face an audience that is interested in information relevant to valuation purposes. Firms are motivated to disclose information that is useful for fulfilling this demand and which complements or substitutes information provided through mandatory disclosure (i.e., intangible assets on the balance sheet). Additionally, due to European regulations, charts used in roadshow and investor conference presentations must be made publicly available. To obtain more specific results, we look at five different intellectual capital categories separately that follow previous research (García-Meca et al., 2005, Bukh et al., 2005): human capital (HC); customer capital (CUS); process capital (PRO); research, development and innovation capital (RDI); and strategy capital (ST). This allows for a detailed analysis of the relationship between recognition of intangible assets and each individual category of intellectual capital. Our study contributes to the literature in three regards. First, we highlight the importance and impact of intangible assets as a determinant of voluntary intellectual capital disclosure. This is relevant for future studies on voluntary intellectual capital disclosure because we show that intangible assets cannot be ignored as a determinant. Thus, capitalised intangible assets must be considered in future voluntary intellectual 3

5 capital disclosure studies to avoid problems of omitted variables. Second, the results of our study are of interest to standard setters. We demonstrate that and how recognition of intangible assets is related to the voluntary intellectual capital disclosure practices of firms. In general, recognition of intangible assets substitutes voluntary disclosure, which indicates that firms value the recognition of intangible assets as being useful for their investors and analysts. However, there are also intellectual capital categories that are positively related to intangible assets, meaning voluntary disclosure about these categories complements the recognition of intangible assets. Third, in additional analyses, we provide detailed evidence on the association between capitalised intangible assets and different intellectual capital categories. This will be useful in additional research aiming to identify for which categories there is a substitution of recognition and voluntary disclosure and for which categories there is no such substitution. Through descriptive statistics and further analysis we reveal practical use of voluntary intellectual capital disclosure of Germany s largest companies in roadshow and investor conference presentations. The remainder of the paper is structured as follows. In the next section, we discuss the research on voluntary intellectual capital disclosure, explain how mandatory disclosure theoretically influences voluntary disclosure, and develop hypotheses. In section 3, we provide the research design. Section 4 shows the data selection. The results for different models and robustness tests are reported in section 5, and section 6 concludes. 4

6 2. Voluntary intellectual capital disclosure, recognition of intangible assets, and private channel communication Ambiguous relationship between voluntary and mandatory disclosure While financial reporting is criticised for providing insufficient information about intellectual capital (Wyatt, 2008), voluntary disclosure enables firms to report additional intellectual capital information and is indeed used for this purpose (García-Meca and Martinez, 2007). Following economics-based theories, this serves the purpose of reducing investor uncertainty regarding a firm s future cash flows (and earnings) and thus reduces the cost of capital (Leuz and Wysocki, 2008). This assumption is extensively tested and supported in the empirical literature (Botosan, 2006). Furthermore, it is empirically shown that the reporting of certain intellectual capital metrics is positively linked to a firm s share price. For example, higher research and development expenditures (Amir and Lev, 1996) and greater customer satisfaction (Ittner and Larcker, 1998) are related to higher share prices. Hence, firms disclose intellectual capital information to inform investors about future (increases in) earnings and cash flows. Analytical papers provide theoretical explanations based on economics-based theories, such as principal-agent theory (Jensen and Meckling, 1976), for disclosure decisions in general. In the absence of any disclosure costs, there is equilibrium where all firms, except the worst performing ones, disclose their information to distinguish themselves from the worst performing firms (Verrecchia, 1983). However, when disclosure costs exist, disclosure is no longer beneficial for all firms. For example, the second-worst performing firm might become the worst performing firm when it chooses to spend additional disclosure costs. Hence, following economics-based theory, firms must weigh the costs and benefits of disclosure, and the equilibrium changes, as, for 5

7 example, Verrecchia (1983) and Dye (1985) analytically show. However, most analytical papers are concerned with the proprietary costs of isolated voluntary disclosure decisions (i.e., voluntary disclosure in the absence of any other related disclosure). This is problematic because, on the one hand, if information is already disclosed mandatorily, then there are no additional benefits gained from voluntarily disclosing such information. Hence, if disclosure costs exist, then mandatory disclosure may substitute voluntary disclosure. On the other hand, a firm might use voluntary disclosure to better explain its mandatory disclosures or to widen the scope of mandatory disclosure. For example, a firm that mandatorily reports capitalised research and development expenditures can use voluntary disclosure to inform the public about successful patents or the impact of newly developed products on sales to show the output of its R&D department. In this case, voluntary disclosure complements mandatory disclosure. Theoretical research that considers the relationships between mandatory and voluntary disclosure (Dye, 1990, Einhorn, 2005, Bagnoli and Watts, 2007) argues that if mandatory disclosure is taken into account, the direction of the relationship between mandatory and voluntary disclosure becomes ambiguous. Whether voluntary disclosure is increased or decreased under the additional consideration of mandatory disclosure depends on various factors. Dye (1990) develops a model in which the association between mandatory and voluntary disclosure depends on several factors, such as the nature of the disclosed information (Does the disclosure have a real effect on other firms, for example, because the revealed information changes market prices for products?); the risk preferences of the firm, its shareholders and outside investors; the mandatory disclosure policy; and the interrelationships between firms (and their cash flows). According to the model of Einhorn (2005), the association between mandatory 6

8 and voluntary disclosure depends on the extent of the correlation between mandatorily disclosed information and private information (i.e., the information that is held by the firm alone and that is not publicly available unless the firm discloses it) as well as on the type of the mandatorily disclosed information and the private information (i.e., does it constitute good or bad news). Additionally, Bagnoli and Watts (2007) take into account whether private information substitutes or complements mandatory disclosure and whether the item about which management possesses private information is permanent or transitory. To summarise, the association between mandatory and voluntary disclosure depends on the specific information content of both voluntary and mandatory disclosure. It is argued that the nature of the externality, good news vs. bad news, permanent information vs. transitory information, and private information vs. public information all play important roles. The analytical papers described above show that there are three possible scenarios for the relationship between mandatory and voluntary intellectual capital disclosure: (1) Increased recognition of intangible assets leads to increased voluntary intellectual capital disclosure; (2) increased recognition of intangible assets leads to decreased voluntary disclosure; and (3) recognition of intangible assets is not connected to voluntary disclosure. Which association exists for the information in question depends on the characteristics of the information, as discussed above. However, to the best knowledge of the authors, no paper exists that comprises all possible combinations of information characteristics, and therefore, the type of association cannot be identified a priori. Furthermore, a field as wide as intellectual capital disclosure covers many different information items, which have different characteristics. Hence, it would be difficult to identify an overall relationship between mandatory and voluntary disclosure, even if analytical research could provide a comprehensive model. 7

9 Disclosure of intangible assets and intellectual capital Intangible assets were rarely included in financial statements in the past, the main reason being that the property rights for controlling intangible assets are unclear (e.g., a company does not own its employees), future earnings derived from intangible assets are uncertain, and the value of intangible assets often relies on unknown context factors (e.g., process capital in the form of a superior workflow may rely on a unique firm culture). Thus, intangible assets are difficult to identify (cf. IAS 38.8). As a result, the financial reporting of intellectual capital is highly asymmetric, meaning that intellectual capital becomes an intangible asset recognised on the balance sheet (e.g., as part of goodwill) when it is purchased (the property rights are clear and the purchase price is an objective estimate of the value of the intangible asset) but not when it is developed internally by the firm itself. This is evident in accounting standards (e.g., in our setting in the German GAAP and, to a lesser extent, in the IFRS and US-GAAP). In the German GAAP, firms could only recognise intangible assets on balance sheets when they were purchased separately, and the recognition of goodwill was optional. [2] Under the IFRS, intangible assets are recognised on the balance sheet whenever they can be identified or have been purchased either separately or as part of a business combination (IAS 38 and IFRS 3). IAS 38.8 requires identifiability, control, and future economic benefits as prerequisites for recognition. Development expenses must be capitalised when certain criteria are fulfilled (IAS 38). [3] Overall, the asymmetric recognition of intellectual capital in financial accounting is evident. Hence, two firms with similar intellectual capital may report fundamentally different values of intangible assets. This may motivate firms that recognise only a small portion of their total intellectual capital on the balance sheet to increase voluntary disclosure. When we apply the three scenarios provided by analytical research (see above) to 8

10 the field of intellectual capital disclosure, the relationship between recognised intangible assets (i.e., mandatory disclosure) and voluntary intellectual capital disclosure can take the following forms. In the first case, voluntary intellectual capital disclosure complements recognition of intangible assets. This is the case when a firm explains or justifies its recognition of intangible assets on the balance sheet. For example, empirical research finds that reported goodwill is value relevant but less so compared to tangible assets (e.g., Barth and Clinch, 1996, Chauvin and Hirschey, 1994). Furthermore, Wyatt (2008) argues that (amortisation of) goodwill is not very reliable. A firm may try to improve the reliability and value relevance of goodwill by voluntarily reporting additional information about the goodwill valuation. In comparison, a firm that created firm value organically cannot report goodwill and hence does not need to explain goodwill accounting. We refer to this scenario as the Complementary Case. H1: Increased recognition of intangible assets leads to increased voluntary intellectual capital disclosure. In the second case, voluntary disclosure is substituted by recognition of intangible assets. This is the case when mandatory disclosure is beneficial for investors and voluntary disclosure can only add marginal benefits. For example, if investors perceive reported goodwill to provide sufficient information, then additional reporting of underlying intellectual capital is unnecessary. However, a firm that created firm value organically, and hence cannot report goodwill, may increase its voluntary intellectual capital disclosure to inform analysts and investors about these investments and the expected cash flows. In this case, the direction of the relationship between recognised intangible assets and voluntary intellectual capital disclosure is the opposite of hypothesis H1. Therefore, we formulate a competing hypothesis for the Substitution Case : 9

11 H2: Increased recognition of intangible assets leads to decreased voluntary intellectual capital disclosure. Note that there is a third scenario in which voluntary disclosure and mandatory disclosure are independent. This is the underlying assumption of most discussions of voluntary disclosure (Leuz and Wysocki, 2008), and is evident in empirical studies examining voluntary disclosure independently of the recognition of intangible assets (Boesso and Kumar, 2007, García-Meca et al., 2005, Prencipe, 2004, Shareef and Davey, 2005). However, there is no empirical evidence regarding this important assumption. This is troublesome because if an association exists between mandatory and voluntary disclosure, ignoring the association leads to an omitted variable problem in empirical research. By addressing this gap in the literature, we lay important groundwork for future empirical studies. Intellectual capital is often categorised into human capital, customer capital and structural capital (Edvinsson and Malone, 1997, Petty and Guthrie, 2000). García-Meca et al. (2005) and Bukh et al. (2005) further distinguish six categories: human capital; customer capital; process capital; research, development and innovation capital; technology capital; and strategic capital. This more detailed categorisation is advantageous for our research question because there are differences in how each of the categories is recognised in financial reporting. For example, for customer capital, concessions may be reported as intangible assets. Research, development and innovation capital may contain capitalised development expenses as well as patents that are recognised as intangible assets. However, for human capital, process capital or strategic capital, the prerequisites for recognition as intangible assets on the balance sheet based on IAS 38.8 are usually not fulfilled. Hence, recognitions only occur as part of goodwill in a business combination. These differences in mandatory disclosure lead to the 10

12 expectation that voluntary disclosure differs among the intellectual capital categories. For example, Bukh et al. (2005) and García-Meca et al. (2005) show that there are more human capital items, customer capital items and strategic capital items disclosed, while there are less items disclosed for process capital, research capital and technology capital. Accordingly, we formulate hypothesis H3: H3: The direction of the relationship between recognition of intangible assets and voluntary intellectual capital disclosure differs across the intellectual capital categories. Private channel communication We base our analyses on a setting of private channel communication. This means that firms directly communicate with a narrow target audience (e.g., through one-on-one meetings, conference calls or presentations to analysts and investors) rather than a public channel communication through which firms address multiple audiences at once (e.g., annual reports or a firm s website). In roadshow and investor conference presentations, firms talk directly to analysts and investors and therefore tailor their presentations to the needs of this target audience. This setting is well suited for our research for the following reasons. First, within the context of private channels, firms face fewer regulations and accordingly have more options for intellectual capital disclosure compared to public channels. Hence, the differences in intellectual capital disclosure between firms are more pronounced. Second, investors and analysts rely on additional information as provided via private channels for firm valuation. For example, Holland (2006) shows that fund managers 11

13 regard information about intellectual capital as highly relevant for firm valuation and request such information via private channel communication. García-Meca et al. (2007) report that analysts request information about intellectual capital, especially if such information is not available via financial reporting. Third, firms are aware of the demand for information about intellectual capital. Günther et al. (2005) find that firms executives perceive intellectual capital to be an important driver of changes in share prices. Accordingly, firms can tailor their voluntary intellectual capital disclosure directly to the needs of analysts and investors in private channel communications. In a similar approach, García-Meca et al. (2005) use the setting of private channel communication to analyse the voluntary disclosure of intellectual capital for a sample of Spanish firms. In our research setting, presentations to analysts and investors can be analysed closely. This is due to regulations on privileged information and insider trading (Directive 2003/6/EC), which, among other things, require firms to make information from such presentations public. While those publications enable research on private channel disclosure, it is likely that the requirement of publications has some influence on the disclosed information. For example, firms will be more careful in formulating future expectations about market developments if those expectations are made public to avoid the proprietary costs of disclosure. However, the primary setting of a private channel disclosure is still existent, as firms can emphasise certain issues that are important but go unrecognised (or not sufficiently recognised) in its financial statements. Overall, the regulation of disclosure in investor conferences and roadshow presentations prevents a false or one-sided disclosure, as management can be prosecuted for such behaviour. This levels the playing field. 12

14 3. Sample We observe firms that were listed in the DAX 30 index in the year To cover a longer window of time, we follow Lopes and de Alencar (2010) and analyse presentations every two years, starting in 2001 and ending in Thus, the timeline ends before distortions stemming from the 2008 financial crisis can affect it. With the focus on the DAX 30 index, we analyse a homogenous sample of the largest firms in Europe s largest economy, meaning firms with an expected high degree of investor relations, international listings and a high level of awareness of communication with analysts. Hence, these firms are likely to intensely consider the costs and benefits of intellectual capital disclosure. We observe that some firms hold more than one roadshow presentation per month, use the same or similar presentations for their presentations and sometimes provide only one set of charts for these presentations. In order to avoid a bias due to different disclosure practices of roadshow presentations (e.g., firms with several similar roadshow presentations per month vs. firms with only one or two roadshow presentations per year), we restrict the number of analysed roadshow presentations to a maximum of one per month. If more than one set of charts is available for a certain month, we analyse the most comprehensive presentation (in terms of the number of slides). In total, 264 presentations are analysed. Table 1 shows how many roadshow and investor conference presentations are analysed in each of the observed years. [Insert Table 1 about here] We derive data for the market value of equity from Thomson Reuters Datastream and necessary accounting data from Worldscope. 4. Research design and descriptive statistics In the preceding section, we outlined the relationship between mandatory and voluntary 13

15 disclosure, explained intellectual capital disclosure, and discussed the characteristics of private channel communication. Now that those issues are clarified, we develop our research design in this section. First, we introduce measures for both voluntary intellectual capital disclosure and recognition of intangible assets. Second, we introduce the econometric models and additional control variables. Voluntary intellectual capital disclosure We identify 93 quantitative intellectual capital disclosure items for the 264 presentations. [4] For the identification, we started with the examples of key performance indicators for intellectual capital as provided by the German working group Accounting and Reporting of Intangibles (WGARI) (Arbeitskreis "Immaterielle Werte im Rechnungswesen" der Schmalenbach-Gesellschaft für Betriebswirtschaft e. V., 2005). In cases in which a presentation contained items that covered intellectual capital but were not mentioned by the WGARI, this item was added to the coding scheme after a discussion between the two coders. Note that we try to capture intellectual capital items in a wide range following our initial definition of intellectual capital (i.e., the entity of a firm s nonphysical claims to future benefits). Therefore, for some intellectual capital items, it is not possible to draw a direct connection to intangible assets (e.g., labour turnover rate), while for other items, this can be done (e.g., the item customer satisfaction may be connected to brand values). Not narrowing the scope of intellectual capital items in our research approach was a deliberate choice because we want to focus on the interplay between the unregulated voluntary disclosure, through which firms can report on any item that they judge to be relevant to analysts and investors, and mandatory disclosure about intangible assets, which is restricted by accounting standards. 14

16 The intellectual capital items are classified into five intellectual capital categories: human capital (HC); customer capital (CUS); process capital (PRO); research, development and innovation capital (RDI); and strategic capital (ST). [5] The list of all 93 identified items with categories and frequencies is given in the appendix. [6] This classification takes into account the common approach of distinguishing intellectual capital from human capital, customer capital and structural capital (Edvinsson and Malone, 1997, Petty and Guthrie, 2000). Furthermore, our classification follows the more detailed categorisation of García-Meca et al. (2005) and Bukh et al. (2005), with one exception. We do not analyse technology separately because this category only contains a few items, as García-Meca et al. (2005) show. Table 2 denotes the number of items identified for each intellectual capital category, and hence comprises the basis of the quantity disclosure score. The quantity disclosure score follows García-Meca et al. (2005) and counts the number of different intellectual capital items reported in one presentation of a firm. Repetitions are not taken into account, as they do not deliver additional information. [Insert Table 2 about here] The calculation of the quality disclosure score is based on the validated voluntary disclosure score of Boesso and Kumar (2007). The score comprises three dimensions: type (qualitative or quantitative disclosure), nature (financial or non-financial disclosure) and outlook (forward- or backward-looking disclosure). We differ from Boesso and Kumar (2007) in one regard: we do not count repeated disclosure of items because they do not add new information. Instead, we analyse how thoroughly each dimension is disclosed throughout the entire presentation. For each dimension and each of the five intellectual capital categories, we apply two points each if the information disclosed is quantitative, non-financial or forward looking. We apply one point each if the 15

17 information disclosed is qualitative, financial or backward looking. We apply no point when there is no information regarding an intellectual capital category. Following this procedure, the theoretical maximum of the quality disclosure score is 30 (a maximum of two points for each of the three dimensions and for each of the five intellectual capital categories). Similar to the measurement of the quantity disclosure score, we measure the quality disclosure score based on the points of each intellectual capital category throughout the entire presentation of a firm. If more than one item is disclosed in one category, the evaluation is based on the maximum values achieved per IC category. The coding was carried out by two researchers. Both researchers developed a similar understanding by analysing three reports together and continuously aligning their coding throughout the content analysis to assure the reliability of the score. One might argue that it would be more appropriate to measure the magnitude of voluntary disclosure. However, voluntary intellectual capital disclosure provides more than just monetary information of intellectual capital investments, and even such monetary information is given in a multitude of different ways, which makes the definition of one magnitude measure impractical. Therefore, we judge our approach to be appropriate for capturing the characteristics of voluntary intellectual capital disclosure with the available information. To sum up, we apply two measures to capture voluntary intellectual capital disclosure (Vol. Disc.): a disclosure quantity score based on García-Meca et al. (2005) and a disclosure quality score based on Boesso and Kumar (2007). Recognition of intangible assets While there is ample literature on which to base the measurement of voluntary intellectual capital disclosure, this is not the case for the recognition of intangible assets. However, Lev and Zarowin (1999) argue that the declining relevance of financial 16

18 reporting, which they report in their analysis, is mainly caused by the negligence of intellectual capital investments on the balance sheet. Accordingly, they propose to more comprehensively capitalise such investments. Our measure of recognition of intangible assets is connected to this proposal; it is the portion of intangible assets recognised on the balance sheet in relation to total assets. The portion increases if a firm capitalises its intellectual capital investments, e.g., a firm capitalises its development expenditures. The portion will not increase if the intellectual capital investment cannot be capitalised or if the firm simply did not invest. Note that in the latter case, the association between recognition of intangible assets and voluntary disclosure has a positive direction: if a firm recognises a small portion of intangible assets on the balance sheet due to small intellectual capital investments, then intellectual capital is less important to the firm, and, accordingly, there will be less voluntary disclosure. This is a shortcoming of our measure. We mitigate possible negative effects by applying industry dummy variables. With this approach, we control for different levels of intellectual capital investments between industries. Sometimes the difference between market and book value of a firm is attributed to the importance of intellectual capital (Lev and Zarowin, 1999, Boesso and Kumar, 2007). Accordingly, one may argue for measuring the recognition of intangible assets as the ratio of capitalised intangible assets to the difference between the market value and book value of a firm. We do not apply such market-based measures in our basic analyses. The first reason is that this difference can also be attributed to another cause: the perceived quality of financial reporting (Tasker, 1998) due to, for example, unrealised gains in the firm s assets. The second reason is that the market value of a firm is rather volatile and influenced by numerous external factors, which would add noise to our measure of recognition of intangible assets. However, in the section on the 17

19 robustness of results, we discuss an alternative measure of recognition of intangible assets based on market value. On the face of the balance sheet, there are two items concerned with recognition of intangible assets: other intangible assets and goodwill. Other intangible assets cover all capitalised investments in intellectual capital that fulfil the criteria of the underlying accounting standard to be recognised as intangible assets. Goodwill is reported for business combinations and is the difference between the purchase price of the acquired firm and the value of all of the assets (including other intangible assets) less the value of the liabilities after the purchase price allocation. Goodwill captures the value of intellectual capital, which cannot be evaluated separately and hence cannot be recognised as individual items on the balance sheet, for example, the additional price that the acquirer pays for the knowledge and creativity of the acquired firm s workforce. An asymmetric recognition of intellectual assets is especially evident for goodwill. A firm that obtains intellectual capital via a business combination reports goodwill, while a firm that invests in creating intellectual capital organically must expense its investments if they cannot be recognised as separate items (e.g., investments in customer relations or human resources). We measure the magnitude of recognition of intangible assets by calculating the portion of other intangible assets and goodwill on total assets (Intangibles&Goodwill). For more detailed analyses, we also measure the portion of other intangible assets on total assets (Intangibles) and goodwill on total assets (Goodwill) separately. Model description and control variables With model 1, we analyse the basic relationship between the recognition of intangible assets and the quality and quantity of the voluntary disclosure of intellectual capital (Vol. Disc.). Following the prior literature on explanatory factors of intellectual 18

20 capital disclosure (García-Meca et al., 2005, Gelb, 2002, Lang and Lundholm, 1993), we control for the influence of firm size (log of total assets), profitability (return of equity), leverage (debt divided by equity), financing of the company (increase in common stocks and/or debt in the fiscal year following the presentation), market-to-book ratio (the market value of equity divided by the book value of equity) and the type of presentation (a dummy variable for which investor conference equals one and roadshow equals zero). Furthermore, we control for the industry, the year of the presentation and the accounting standard applied in the fiscal year preceding the presentation. We acknowledge the panel structure of our sample by applying Tobit regressions with two-way clustered standard errors (Petersen, 2009, Gow et al., 2010). We cluster by firms and the month of the presentation. The two-way clustering considers that residuals are correlated across firms and across months of the presentations and corrects standard errors accordingly. Gow et al. (2010) argue and show that two-way clustering is the most suitable approach when confronted with both cross-sectional and time-series dependencies. [7] We apply Tobit regressions to acknowledge that the dependent variable is not continuous but is censored on one side, which leads to a higher number of observations at (or close to) the censor value. In our case some reports do not contain any information about intellectual capital, and hence, the disclosure quality and quantity are censored at zero. Model 1: Vol. Disc. = β 0 + β 1 Intangible Assets and Goodwill + β 2 Size + β 3 Profit + β 4 Leverage + β 5 Financing + β 6 Market-to-book ratio + β 7 Conference + controls industry + controls year + controls accounting + ε 1 In compliance with prior literature, we expect the sign of firm size (Size) to be 19

21 positive (Adams et al., 1998, García-Meca et al., 2005, Lang and Lundholm, 1993, Patten, 1991), meaning that larger firms provide more voluntary disclosure. Higher profits (Profit) (García-Meca et al., 2005, Lang and Lundholm, 1993) as well as a higher financial leverage (Leverage) (Jason et al., 1995, Prencipe, 2004) should also lead to more voluntary disclosure. If a firm plans to raise capital, it will increase its disclosure beforehand (Frankel et al., 1995), leading to a positive relationship between planned financing (Financing) and voluntary disclosure. The Market-to-book ratio is used in different contexts in disclosure studies. At first, the market-to-book ratio is interpreted as an estimator of the perceived quality of prior public information (Frankel et al., 1999, García-Meca et al., 2005, Tasker, 1998). If the market-to-book ratio is high, then the perceived quality of financial reporting is low, which means the book value only covers a small portion of the firm s total market value. This motivates firms with a high marketto-book ratio to increase their voluntary disclosure. Second, the market-to-book ratio is sometimes interpreted as the relevance of intellectual capital to a firm (Boesso and Kumar, 2007, Lev and Zarowin, 1999). Hence, firms with a high market-to-book ratio are expected to have intellectual capital of a high value and, accordingly, have more motivation to disclose intellectual capital information voluntarily. However, the difference between market and book values can be claimed to be due to more than just intellectual capital. For example, tangible assets may be reported with unrealistically low values on the balance sheet (hidden reserves) or financial market expectations lead to volatile market prices (Boesso and Kumar, 2007, Brennan, 2001). Both interpretations of the market-to-book ratio support a positive relationship with voluntary intellectual capital disclosure. The variable Conference is equal to one if the presentation is an investor conference and equal to zero otherwise. Investor conference presentations are usually longer and more comprehensive than roadshow presentations. Thus, we expect 20

22 the sign of the coefficient for the dummy variable Conference to be positive (García- Meca et al., 2005). Furthermore, we apply control variables for different industries (controls industry ), years (controls year ) and accounting standards (controls accounting ). The latter captures whether the accounting standard applied in the financial year ending prior to the presentation is German GAAP, US-GAAP, or IFRS [8]. The definitions and calculations of all of the variables are summarised in Table 3. [Insert Table 3 about here] For model 2, we decompose recognition of intangible assets into two components: reported goodwill and other intangible assets. We now apply two measures of recognition of intangible assets. The first measure captures the portion of reported goodwill on total assets (Goodwill), and the second measure captures the portion of other intangible assets on total assets (Intangible assets). Goodwill and Intangible assets have a different nature in the sense of symmetric recognition of their underlying intangible assets. As explained above, Goodwill leads to extremely asymmetric recognition, meaning that in a business combination, goodwill is recognised as a result of purchase price allocation. However, investments with the goal of internally generating goodwill by organic growth are expensed. Intangible assets are more symmetric than Goodwill. For example, IAS 38.8 requires identifiability, control, and future economic benefits as prerequisites for recognition as intangible assets. These criteria can be fulfilled by both self-created and purchased intellectual capital. Due to the different nature of the two balance sheet items, Goodwill and Intangible assets may have different relationships with voluntary disclosure, which is acknowledged in model 2. All other variables are the same as in model 1. 21

23 Model 2: Vol. Disc. = β 0 + β 1 Intangible Assets + β 2 Goodwill + β 3 Size + β 4 Profit + β 5 Leverage + β 6 Financing+ β 7 Market-to-book ratio + β 8 Conference + controls industry + controls year + controls accounting + ε 2 As discussed in section 2, it is likely that intellectual capital categories are associated differently with the recognition of intangible assets. For example, while there are specific regulations about the capitalisation of research and development expenses in the IFRS (IAS 38), human capital cannot be recognised on the balance sheet, except as part of goodwill. Therefore, we formulated hypothesis H3 (different directions across IC categories of the relationship between recognised intangible assets and the voluntary intellectual capital disclosure). For this analysis, we apply multivariate regressions to analyse the relationship between reported goodwill and other intangible assets on the one side and voluntary disclosure regarding each of the five intellectual capital categories on the other side: human capital (Disc.HC); customer capital (Disc.CUS); process capital (Disc.PRO); research, development and innovation capital (Disc.RDI); and strategic capital (Disc.ST). We apply multivariate regressions in order to consider that disclosure decisions for all five intellectual capital categories are made simultaneously: Model 3:

24 = β 0 + β 1 Intangible Assets + β 2 Goodwill + β 3 Size + β 4 Profit + β 5 Leverage + β 6 Financing+ β 7 Market-to-book ratio + β 8 Conference + controls industry + controls year + controls accounting + ε 3 Descriptive statistics The descriptive statistics of the dependent variables and independent variables are summarised in Table 4. On average, the disclosure quality score is relatively low at 8.432, considering that the theoretical maximum is 30. Additionally, the disclosure quantity score is relatively low. On average, we find different reported intellectual capital items per presentation. Both statistics show that firms choose their intellectual capital disclosure very specifically. For intellectual capital categories, we find the highest disclosure quality scores for customer capital (2.875) and human capital (1.598), and the highest disclosure quantity scores for strategy capital (0.886) and customer capital (0.780). [Insert Table 4 about here] Table 5 reports the correlation coefficients of disclosure quantity, disclosure quality and independent variables. The disclosure quantity and disclosure quality are strongly correlated, and as one would expect, both goodwill and other intangible assets are strongly correlated with total intangible assets. All of the other correlations seem rather small and - except for Profit and the Market-to-book ratio - do not exceed 0.5, which indicates that problems of multicollinearity do not exist. [9] The correlations table reports a negative association between Intangible assets & goodwill (and Intangible assets) and both disclosure scores. However, most of these associations are not 23

25 significant and thus need further exploration. [Insert Table 5 about here] 5. Results Analyses of disclosure quantity and quality The results of model 1 presented in Table 6 strongly support a relationship between the intangible assets recognised on the balance sheet and the voluntary intellectual capital disclosure quantity (coefficient for Intangible assets and goodwill ), as well as quality (coefficient ). The relationship shows a negative sign, which supports hypothesis H2, meaning that voluntary intellectual capital disclosure is substituted by the recognition of intangible assets. When decomposing Intangible assets and goodwill into the two variables Goodwill and Intangible assets (model 2), it becomes evident that this negative relationship is attributable to Goodwill alone. If the amount of goodwill reported on the balance sheet is high, then a firm expects few benefits from additional voluntary disclosure. Interestingly, neither H1 nor H2 are supported for Intangible assets because we find no significant relationship between this variable and voluntary intellectual capital disclosure. However, if the intellectual capital categories show different relationships with Intangible assets, then it would be inappropriate to analyse only the total quantity and quality of disclosure. Hence, we additionally analyse each individual intellectual capital category in the following section. [Insert Table 6 about here] Further results show that the quantity of voluntary intellectual capital disclosure is not 24

26 related to any of the other explanatory factors suggested by the prior literature, except for the type of meeting. The variable Conference shows a significant coefficient and indicates that investor conference presentations contain more intellectual capital disclosures than roadshows. This is as expected because investor conferences are longer and provide more detailed information. [10] The quality of intellectual capital disclosure is negatively related to Profit. Hence, in the face of poor earnings, firms tend to report more about their intellectual capital investments in order to justify poor earnings and assert that there is potential for future success, leading to higher subsequent earnings. Analyses of disclosure quantity and quality for individual IC categories Table 7 comprises the results for hypothesis H3 and for each of the five intellectual capital categories. Specifically, we focus on the question of how the dependent variables (disclosure quantity and disclosure quality of each intellectual capital category) are related to Goodwill and Intangible assets. The detailed analyses support H3 and reveal different relationships. For human capital, the substitution case is supported because we find significantly negative relationships with Goodwill and Intangible assets. That means, voluntary disclosure is substituted by recognition of intangible assets. Note that human capital cannot be recognised as a separate intangible asset on the balance sheet. However, it is likely that other intangible assets, for example, the value of patents or software, are at least partly the results of human capital. Customer capital is related significantly neither to Goodwill nor to Intangible assets. However, when a firm plans to raise capital, it also increases its voluntary disclosure about customer capital, both in quantity and quality. 25

27 The quantity of voluntary disclosure of process capital is related negatively to Goodwill (substitution case) and positively to Intangible assets (complementary case). The quality of voluntary disclosure of process capital is only positively linked to Intangible assets (complementary case). However, the relationship is weaker compared with human capital regarding the coefficients and the significance levels. For research, development, and innovation capital, we find support for the substitution case for Goodwill (regarding both, disclosure quality and quantity) and support the complementary case for Intangible assets and disclosure quality. For the quality of IC disclosure the positive relationship between process capital as well as research, development, and innovation capital, and Intangible assets is noteworthy. Here, firms provide additional information when there is a larger proportion of intangible assets recognised on the balance sheet. In this case, voluntary disclosure serves to complement financial reporting data. For example, as capitalising development expenditures under IFRS is a relatively new accounting procedure, firms might explain such capitalisations more thoroughly. However, higher goodwill is significantly related to lower disclosure quantity and to quality of research, development, and innovation capital. Finally, the disclosure quantity of strategic capital is significantly and negatively related to Goodwill (substitution case) while the disclosure quality of strategic capital is significantly and negatively related to Intangible assets (complementary case). That means goodwill reporting substitutes voluntary disclosure of strategic capital because goodwill provides a financial measure of the prospects that are captured in strategic capital, particularly concerning synergies and going concern (Johnson and Petrone, 1998). 26